1:
2: % Template article for preprint document class `elsart'
3: % SP 2001/01/05
4:
5: \documentclass{elsart}
6:
7: % Use the option doublespacing or reviewcopy to obtain double line spacing
8: % \documentclass[doublespacing]{elsart}
9:
10: % if you use PostScript figures in your article
11: % use the graphics package for simple commands
12: % \usepackage{graphics}
13: % or use the graphicx package for more complicated commands
14: \usepackage{graphicx}
15: % or use the epsfig package if you prefer to use the old commands
16: % \usepackage{epsfig}
17:
18: % The amssymb package provides various useful mathematical symbols
19: \usepackage{amssymb}
20:
21: \begin{document}
22:
23: \begin{frontmatter}
24:
25: % Title, authors and addresses
26:
27: % use the thanksref command within \title, \author or \address for footnotes;
28: % use the corauthref command within \author for corresponding author footnotes;
29: % use the ead command for the email address,
30: % and the form \ead[url] for the home page:
31: % \title{Title\thanksref{label1}}
32: % \thanks[label1]{}
33: % \author{Name\corauthref{cor1}\thanksref{label2}}
34: % \ead{email address}
35: % \ead[url]{home page}
36: % \thanks[label2]{}
37: % \corauth[cor1]{}
38: % \address{Address\thanksref{label3}}
39: % \thanks[label3]{}
40:
41: \title{Nuclear Structure, Random Interactions and Mesoscopic
42: Physics}
43:
44: % use optional labels to link authors explicitly to addresses:
45: % \author[label1,label2]{}
46: % \address[label1]{}
47: % \address[label2]{}
48:
49: \author[nscl]{Vladimir Zelevinsky}
50: \author[fsu]{Alexander Volya}
51: \address[nscl]{National Superconducting Cyclotron Laboratory and
52: Department of Physics and Astronomy, Michigan State
53: University, East Lansing, MI 48824-1321, USA}
54: \address[fsu]{Department of Physics, Florida State University, Tallahassee,
55: FL 32306-4350, USA}
56:
57: \begin{abstract}
58: Standard concepts of nuclear physics explaining the systematics of
59: ground state spins in nuclei by the presence of specific coherent
60: terms in the nucleon-nucleon interaction were put in doubt by the
61: observation that these systematics can be reproduced with high
62: probability by randomly chosen rotationally invariant
63: interactions. We review the recent development in this area, along
64: with new original results of the authors. The self-organizing role
65: of geometry in a finite mesoscopic system explains the main
66: observed features in terms of the created mean field and
67: correlations that are considered in analogy to the random phase
68: approximation.
69: \end{abstract}
70:
71: \begin{keyword}
72: % keywords here, in the form: keyword \sep keyword
73:
74: % PACS codes here, in the form: \PACS code \sep code
75: \PACS
76: \end{keyword}
77: \end{frontmatter}
78:
79: \section{Introduction}
80:
81: It became a common place to claim that the basic facts of the
82: nuclear ground state systematics, namely that all even-even nuclei
83: have ground state spin $J_{0}=0$ and the lowest possible isospin
84: $T_{0}=T_{{\rm min}}$, are due to the fundamental properties of
85: residual nucleon-nucleon forces. The {\sl pairing} phenomenon was
86: known, in particular through the mass formula, from the beginning
87: of nuclear spectroscopy. It was formulated by Racah in an elegant
88: form with the use of the seniority quasispin formalism
89: \cite{Racah,Talmi}. The predictions of the simple shell model by
90: Mayer and Jensen \cite{MJ} would be uncertain for all nuclei,
91: except magic ones and those with one particle or one hole on top
92: of the magic core, if the pairing would not allow one to guess
93: that the nuclear ground state spin in odd-$A$ nuclei is determined
94: by the last unpaired nucleon. A. Bohr, Mottelson and Pines
95: \cite{BMP} found a profound analogy of nuclear pairing to
96: superconducting pair correlations of electrons in metals, and
97: Belyaev in his seminal paper \cite{Bel} demonstrated how the
98: pairing interaction creates correlations that modify not only the
99: ground state energy but all single-particle and collective
100: properties of low-lying nuclear states, in a good agreement with
101: observed trends \cite{KS}.
102:
103: Following the advances in experiments and theory towards nuclei
104: far from stability, the interest to pairing has recently
105: increased. The pairing correlations play a decisive role in
106: determining the nuclear drip-line; many nuclides can be stable
107: only due to the pairing correlations between the outermost loosely
108: bound nucleons. The pairing interaction is an inalienable and very
109: important part of all modern shell model versions \cite{BAB}.
110:
111: Therefore an unexpected observation by Johnson, Bertsch and Dean
112: \cite{JBD} was met with great interest and immediately put under
113: the microscope of various tests. They noticed that even {\sl
114: randomly} taken two-body forces acting between the fermions in a
115: restricted Hilbert space of few single-particle orbitals lead to
116: the statistical predominance of the ground state spin $J_{0}=0$.
117: The broad discussion that followed this discovery revealed that
118: similar phenomena take place for interacting bosons as well. The
119: natural questions arise: do we understand well the physics
120: generated by random interactions under constraints of rotational
121: symmetry and what is the reason for empirical regularities in
122: nuclei? The problem is not limited to nuclear physics. Atomic
123: clusters, particles in the traps, quantum dots and disordered
124: systems, such as quantum spin glasses, are just a few examples
125: where the same questions are to be answered.
126:
127: From a more general theoretical viewpoint related to {\sl
128: many-body quantum chaos} \cite{grib,big,ann}, we deal with closed
129: mesoscopic systems that generically display chaotization of motion
130: due to intrinsic interactions. In the absence of a heat bath and
131: external disorder, the interactions play the role of a randomizing
132: or thermalizing factor. They create a very complicated structure
133: of the eigenstates. However, the presence of exact symmetries
134: (rotational, time-reversal, parity, isospin) leads to
135: non-vanishing correlations between the classes of states with
136: different exact quantum numbers since they are governed by a
137: single deterministic (even if randomly picked) Hamiltonian with a
138: relatively small number of parameters. Such correlations bring a
139: new, hardly discussed before, element to theory of quantum chaos.
140:
141: Below we describe the problem more in detail following the main
142: ideas proposed for explanation by various authors. We show that a
143: conventional notion of a mean field created by the interaction
144: removes the main puzzling features of the problem and puts the
145: whole story on a clear track. Of course, the open questions still
146: remain leaving the room for future exciting studies.
147:
148: \section{Two-body interactions in an isolated many-body system
149: and many-body quantum chaos}
150: \subsection{Hamiltonian}
151: Our starting point is a standard {\sl shell-model approach} to a
152: many-body problem. $N$ particles are interacting through two-body
153: forces within Hilbert space that is built on a certain number of
154: single-particle orbitals. We label the single-particle states as
155: $|1)$, incorporating all necessary quantum numbers in the unified
156: label {\sl 1}. Using the single-particle basis diagonalizing the
157: independent particle part, the general Hamiltonian of the system
158: can be written as
159: \begin{equation}
160: H=\sum_{1}\epsilon_{1}a^{\dagger}_{1}a_{1}+\frac{1}{4} \sum_{1234}
161: V(12;34)a^{\dagger}_{1}a^{\dagger}_{2}a_{3}a_{4}, \label{2.1}
162: \end{equation}
163: where we introduced the creation and annihilation operators with
164: usual commutation (anticommutation) rules for bosons (fermions).
165: The interaction matrix elements $V(12;34)$ are correspondingly
166: symmetrized (antisymmetrized) with respect to permutations
167: $1\leftrightarrow 2$ and $3\leftrightarrow 4$. Assuming
168: time-reversal invariance, we can consider all matrix elements
169: real. The restriction to two-body forces (``rank" of the
170: interaction $r=2$) is not significant as long as the total
171: particle number $N\gg r$, and can be removed. Note that the
172: general form (\ref{2.1}) does not explicitly carry any
173: conservation law except for the particle number. Later we consider
174: the requirements of rotational (or isospin) invariance.
175:
176: Two physical formulations can be considered in parallel. In
177: application to a realistic system, the Hamiltonian is derived from
178: more general theory (for example, for nuclei it can be based on
179: meson theory or quark models) or built empirically with the
180: parameters, $\epsilon_{1}$ and $V(12;34)$, adjusted to
181: experimental data. One can also consider {\sl ensembles} of
182: Hamiltonians that satisfy the requirements of Hermiticity and
183: quantum statistics but the parameters, or part of them, are
184: treated as random variables taken from some distribution. The
185: explicitly introduced randomness of the Hamiltonian keeping the
186: same form as that of actual mesoscopic systems was used with the
187: purpose to bring the {\sl global} description in terms of random
188: matrices near to physical reality. It turns out that the {\sl
189: local} spectral statistics, starting from sufficiently high level
190: density, are universal. They express generic properties of
191: many-body quantum chaos (plus the assumption of time reversal
192: invariance) and do not depend on the details of the interaction.
193: Below we first characterize these universal features and then turn
194: to the ground state problem.
195:
196: \subsection{Ensemble of random interactions}
197:
198: The studies of the random two-body interactions go back to
199: Wigner-Dyson {\sl random matrix theory}, see \cite{Fluc} and
200: references therein. This stage of development was thoroughly
201: reviewed by Brody {\sl et al.} in Ref. \cite{Brody}; see also the
202: latest review \cite{Guhr}. The {\sl two-body random ensemble},
203: TBRE \cite{TBRE,TBREa}, in contrast to full canonical (Gaussian
204: Orthogonal, GOE, or Gaussian Unitary, GUE) ensembles \cite{Mehta},
205: considers matrices in many-body Hilbert space, where nonzero
206: off-diagonal matrix elements link the independent particle states
207: that can be connected by two-body processes but not constrained by
208: any conservation laws, except for the symmetry dictated by the
209: particle statistics. These matrix elements are taken in the TBRE
210: as uncorrelated and normally distributed real random quantities.
211: More general {\sl embedded ensembles} \cite{Kota} can be
212: considered with $r$-body forces \cite{embe} for $r<N$; the case
213: $r=N$ with a simultaneous interaction of all particles returns to
214: the full GOE. The angular momentum conservation was, as a rule,
215: ignored because of severe mathematical difficulties \cite{Brody}.
216:
217: We see essential new properties of the TBRE as compared to the
218: canonical random matrix ensembles. (i) The orthogonal, or unitary,
219: invariance of the statistical distribution of random matrix
220: elements is lost. (ii) The natural basis is that of independent
221: particle configurations where only configurations that differ by
222: not more than the occupancies of a pair of orbitals in the initial
223: and final states can be connected by a single-step interaction. In
224: this basis the many-body Hamiltonian matrix is sparse. (iii) The
225: nonzero matrix elements of this matrix are strongly correlated.
226: Indeed, a given two-body scattering process may occur on the
227: background of many different spectator configurations of remaining
228: particles; all many-body matrix elements in those cases are equal
229: regardless of a random or deterministic character of the two-body
230: interaction.
231:
232: Before the work \cite{JBD}, the studies of the TBRE did not
233: consider the consequences of rotational invariance so that the
234: single-particle levels did not carry any additional quantum
235: numbers being fully characterized by their energy. The most
236: important conclusion of these studies was the {\sl ``chaotic"}
237: character of local spectral statistics, essentially the same as
238: predicted for the GOE in spite of a very different distribution of
239: many-body matrix elements. Many results are insensitive to the
240: exact form of the distribution function of random two-body matrix
241: elements that can differ from the Gaussian still remaining
242: symmetric with respect to the sign. It was established \cite{mon}
243: that the global (``secular") behavior of the level density in the
244: given finite Hilbert space of a certain number of single-particle
245: orbitals is close to Gaussian for $N>r$ while for $N=r$ it tends
246: to the semicircle typical for the GOE or GUE. There are only few
247: analytical results for the TBRE and its modifications \cite{Kota},
248: although significant numerical work has been done.
249:
250: \subsection{Complexity of many-body states}
251:
252: The detailed shell model studies for complex atoms \cite{grib} and
253: nuclei \cite{big} showed that realistic forces, Coulomb for atoms
254: and semiempirical effective nucleon-nucleon interactions for
255: nuclei, generate the local spectral statistics well described by
256: the GOE and TBRE within each class of many-body states with fixed
257: exact quantum numbers. Considering the dependence on the
258: interaction strength, the chaotic statistics of nearest level
259: spacings and the so-called $\Delta_{3}$ statistics of level number
260: fluctuations emerge when the interparticle interactions are turned
261: on with their strength still much weaker than the realistic value.
262: This happens without any randomness in the Hamiltonian, in spite
263: of correlations due to the two-body character of the forces and
264: the fact that the realistic distributions of the many-body matrix
265: elements are generically close to exponential rather than Gaussian
266: \cite{big}. The mechanism of spectral chaotization is provided by
267: {\sl multiple avoided crossings} of levels inside a fixed symmetry
268: class.
269:
270: As the interaction strength increases beyond threshold for onset
271: of spectral chaos, and level dynamics with less frequent crossings
272: loses its turbulent character, the main ongoing process is the
273: growth of complexity of the eigenfunctions. The important question
274: here is how one can quantify the {\sl degree of complexity} of an
275: {\sl individual} wave function. The specification of a wave
276: function is always related to a certain basis $|k\rangle$. In the
277: eigenbasis of the Hamiltonian, each eigenfunction has just one
278: component that obviously indicates the absence of complexity. In
279: the above mentioned process of switching on interaction, it is
280: natural to refer all eigenstates to the original basis of
281: noninteracting particles and follow the gradual increase of
282: complexity measured by the number of significant components in the
283: wave function. This choice of the reference basis is also singled
284: out by many-body physics. In a realistic system of the type we are
285: interested in, the single-particle structure is determined by the
286: self-consistent field due to all particles. The mean field
287: embodies the most regular effects of the interaction. The residual
288: interactions already do not contain such average components.
289: Therefore one can think of the mean field basis as the best choice
290: for {\sl separating} regular and chaotic aspects brought in by the
291: interaction \cite{mf}.
292:
293: The degree of complexity of an eigenstate $|\alpha\rangle$ with
294: respect to the reference basis $|k\rangle$ can be quantified with
295: the help of Shannon {\sl information entropy} \cite{izr,entr,big}.
296: If the eigenfunction is given by the normalized superposition
297: \begin{equation}
298: |\alpha\rangle=\sum_{k}C^{\alpha}_{k}|k\rangle, \label{2.2}
299: \end{equation}
300: information entropy of the state $|\alpha\rangle$ in the basis
301: $|k\rangle$ is defined in terms of the weights $w^{\alpha}_{k}=
302: |C^{\alpha}_{k}|^{2}$ as
303: \begin{equation}
304: I_{\alpha}=-\sum_{k}w^{\alpha}_{k}\ln w^{\alpha}_{k}. \label{2.3}
305: \end{equation}
306: As the interaction strength increases, information entropy grows
307: from zero in principle being able to reach the limit of $I_{{\rm
308: max}}=\ln d$, where $d$ is the space dimension. This maximum
309: possible value can be realized for the fully delocalized function
310: with all equal weights $w^{\alpha}_{k}=1/d$. In the GOE the
311: average value of $I_{\alpha}$ is lower than this limit, $I_{{\rm
312: GOE}}=\ln (0.48 d)$, because of the requirements of orthogonality
313: of different eigenstates.
314:
315: The shell model analysis \cite{entr,big} shows that information
316: entropy in all symmetry classes grows smoothly with the interaction
317: strength and in the middle of the spectrum gets close to the
318: $I_{{\rm GOE}}$. With the interaction strength artificially
319: increased beyond its realistic value, one can reach the GOE limit
320: uniformly in excitation energy \cite{temp,big}. It is important
321: that for the realistic, and therefore consistent with the mean
322: field, interaction strength information entropy $I_{\alpha}$ is a
323: smooth monotonously increasing to the middle of the spectrum
324: function of excitation energy $E_{\alpha}$. This allows one to
325: treat information entropy as a {\sl thermodynamic variable} and
326: build up the corresponding temperature scale \cite{temp,big,ann}
327: avoiding any reference to a heat bath or Gibbs ensemble. Thus, one
328: can consider thermodynamics of a closed mesoscopic system based on
329: typical properties of individual quantum states.
330:
331: The physical foundation for that is given by the chaotic mixing of
332: states as a result of the strong interaction at a high level
333: density. This mixing makes statistical properties of closely
334: located states {\sl uniform} (thereby the question by Percival
335: \cite{perc} on a generic relation between the complicated
336: neighboring states is solved - ``the states look the same") and
337: guarantees that macroscopic observables do not depend on the exact
338: population of adjacent microscopic states and the corresponding
339: phase relationships. This is exactly what is needed for the
340: statistical description. Such considerations shed new light on a
341: problem of justification of the thermodynamic approach for closed
342: mesoscopic systems.
343:
344: The quantity $d_{\alpha}=\exp(I_{\alpha})/0.48$ can be interpreted
345: as an effective number of significant (``principal") components of
346: the wave function, or its localization length. The components
347: $C_{k}^{\alpha}$ of a complicated wave function $|\alpha\rangle$
348: on average are uniformly distributed over a sphere of dimension
349: $d_{\alpha}$. The fully uniform distribution on a $d$-dimensional
350: sphere is restricted only by the normalization,
351: \begin{equation}
352: P_{d}(C_{1},...,C_{d})=\frac{\Gamma(d/2)}{\pi^{d/2}}\,
353: \delta(\sum_{k=1}^{d}w_{k}-1), \label{2.4}
354: \end{equation}
355: which leads to the distribution function of any chosen component
356: $C$
357: \begin{equation}
358: P_{1}(C)=\frac{\Gamma(d/2)}{\sqrt{\pi}\Gamma((d-1)/2)}\,
359: (1-w)^{(d-3)/2}, \label{2.5}
360: \end{equation}
361: where, as in eq. (\ref{2.4}), $w=C^{2}$. In the asymptotic limit
362: of large $d$, this distribution goes to the Gaussian. The square
363: of the amplitude has a distribution
364: \begin{equation}
365: P_{1}(w)=\frac{\Gamma(d/2)}{\sqrt{\pi}\Gamma((d-1)/2)}\,
366: \frac{1}{\sqrt{w}}\,(1-w)^{(d-3)/2} \label{2.6}
367: \end{equation}
368: that goes to the Porter-Thomas ($\chi^{2}$) distribution for large
369: $d$. The realistic distributions of the components in the nuclear
370: shell model \cite{verbaar,big}, except for the lowest and the
371: highest states, are close to these predictions with local values
372: of $d_{\alpha}$ smoothly changing along the spectrum, The strength
373: distribution (\ref{2.6}) along with the nearest level spacing
374: distribution can serve as an experimental means for recovering the
375: strength missing in the background of experiments that cannot
376: resolve the invisible fine structure \cite{kilg}.
377:
378: The complexity measure $I_{\alpha}$ or $d_{\alpha}$ gives a tool
379: for estimating matrix elements of simple operators between a
380: simple and complicated state or between two complicated states. In
381: both cases, the typical reduction of the matrix element compared
382: to that between two simple (let say, noninteracting) states is
383: given by the factor $1/\sqrt{d}$ if one assumes that the two
384: complex states have a similar degree of complexity. Since the
385: corresponding level density, which determines energy denominators,
386: increases on average $\propto d$, we come to the {\sl statistical
387: enhancement} of perturbations, $\propto \sqrt{d}$, in the region
388: of many-body quantum chaos \cite{SF}. In light nuclei this
389: enhancement can be seen directly in shell model calculations
390: \cite{AB}. Remarkable examples are given by the strong enhancement
391: of weak interactions in nuclear neutron resonances (parity
392: violation in polarized neutron scattering \cite{Alfim,Bowm,Mitch}
393: and fragment asymmetry in fission by polarized neutrons
394: \cite{Dan,Petr}). Here again we see that statistical regularities
395: in a mesoscopic system coexist with the opportunity to reveal,
396: both theoretically and experimentally, properties of individual
397: quantum states.
398:
399: \subsection{Chaos and thermalization}
400:
401: Another aspect of the same problem is the possibility to
402: describe complicated eigenstates in the standard statistical
403: language of single-particle occupation numbers. It was noticed,
404: both for atoms \cite{grib}, and nuclei \cite{temp,big,ann}, that
405: expectation values of the occupation numbers $n_{j}$ of the mean
406: field orbitals are close to what would be predicted by Fermi-Dirac
407: statistics. Effective temperatures $T_{\alpha}$ extracted for
408: individual states $|\alpha\rangle$ are in good correspondence with
409: thermodynamic temperature determined by the level density as well
410: as with the information temperature found from Shannon entropy.
411: This shows that one can successfully use the notions of {\sl
412: Fermi-liquid theory} modeling the system as a gas of
413: quasiparticles not only near the ground state, as it is usually
414: assumed (in nuclei just in this region the description has to be
415: modified because of pairing correlations \cite{ZVYad}), but
416: practically at any excitation energy below decay threshold. The
417: finite lifetime of quasiparticles is simply translated into
418: statistical occupation factors different from 0 and 1 and smoothly
419: changing along the spectrum. The analytical description of the
420: process of equilibration was given by Flambaum, Izrailev and
421: Casati \cite{FIC}, and Flambaum and Izrailev \cite{FI} in the
422: framework of the TBRE.
423:
424: We need again to stress that information entropy is capable to
425: characterize the degree of complexity only relative to a reference
426: basis. This can be considered as an advantage of information
427: entropy as a measuring tool since we are able to discover
428: relations between the eigenbasis and {\sl various} reference
429: choices. The special role of the {\sl self-consistent} mean field
430: basis is now seen in the consideration of the occupation numbers
431: defined with respect to this basis that forms a skeleton
432: supporting all complications induced by the interaction. The
433: equilibrating factor is the interparticle interaction rather than
434: a heat bath. For the interaction of rigid spheres, which is known
435: to generate chaotic dynamics, it was shown rigorously \cite{mark}
436: that the equilibrium momentum distribution is that of Boltzmann,
437: Bose-Einstein or Fermi-Dirac depending on the statistics of
438: particles even if the interaction cannot be reduced to gaseous
439: rare collisions. In realistic cases the interaction strength is in
440: accordance with the parameters of the mean field. In the case of
441: artificially enhanced interaction, all states go to the GOE limit
442: of complexity, and the single-particle thermometer is not capable
443: of resolving the spectral evolution \cite{temp,big}.
444:
445: The description with the aid of information entropy does not take
446: into account any phase correlations between the components of an
447: eigenfunction. In a sense it gives a {\sl delocalization} measure
448: \cite{big,Kota} of the given state in the original basis of
449: noninteracting particles. It cannot distinguish between an
450: incoherently mixed chaotic state and collective state that is a
451: regular superposition of many basis states with certain phase
452: relationships. The information approach may be also inadequate for
453: an unstable mean field or a phase transition occurring at some
454: temperature (excitation energy). Here another way of
455: characterizing the individual quantum states may be useful
456: \cite{Sok}. One can look at the response of a given state
457: $|\alpha\rangle$ to external noise described by random parameters
458: $\lambda$ in the Hamiltonian. The averaging over $\lambda$
459: determines the density matrix
460: \begin{equation}
461: \rho^{\alpha}_{kk'}=\langle C^{\alpha}_{k}(\lambda)
462: C^{\alpha\ast}_{k'}(\lambda)\rangle_{{\rm av}} \label{2.7}
463: \end{equation}
464: that allows one to define von Neumann entropy
465: \begin{equation}
466: S_{\alpha}=- {\rm Tr}\{\rho^{\alpha}\ln(\rho^{\alpha})\}.
467: \label{2.8}
468: \end{equation}
469: In contradistinction to information entropy, this quantity, that
470: may be called {\sl invariant correlational entropy} (ICE), does
471: not depend on the choice of representation and takes into account
472: correlations between the amplitudes of the wave function.
473:
474: The ICE is very sensitive to quantum phase transitions. If the
475: random parameter $\lambda$ fluctuates around the phase transition
476: point, the strong variation of the structure of the state gives
477: rise to a peak in the ICE as was shown in the interacting boson
478: model (IBM) \cite{cej} and in the realistic shell model
479: \cite{ICE}. One can notice also a common physical aspects shared
480: by the ICE and the notion of {\sl fidelity} extensively studied
481: recently in considerations of quantum dynamics related to quantum
482: echo, decoherence, Zeno effect and quantum computing, see for
483: example \cite{Peres,Niels,Pros}.
484:
485: \section{Rotational invariance}
486: \subsection{Role of symmetries}
487: From the very beginning of studies of random matrices and quantum
488: chaos, the crucial influence of {\sl global symmetries} was
489: repeatedly stressed by many authors. Random matrix ensembles make
490: averaging over all Hamiltonians in a given universality class
491: \cite{Fluc}. The classes are fully determined by the fundamental
492: symmetries as Hermiticity, both for GOE and GUE, and time-reversal
493: invariance (for GOE). The additional requirement for the canonical
494: Gaussian ensembles is the invariance of the distribution of matrix
495: elements under orthogonal or unitary basis transformations. This
496: last demand expresses the limiting property of extreme chaos and
497: brings at our disposal the necessary reference point, against
498: which we can look at the realistic systems with their specific
499: deviations from this limit. The ensembles as TBRE do not obey this
500: requirement but this does not influence the local spectral
501: statistics. In addition an exact permutational symmetry for
502: fermions or bosons is also imposed here.
503:
504: Self-sustaining mesoscopic systems reveal other exact symmetries,
505: first of all rotational symmetry (in the absence of external
506: fields). As a result, any eigenstate is a member of a degenerate
507: rotational multiplet $|JM\rangle$ with total spin $J$ and its
508: projection on the laboratory quantization axis $J_{z}=M$. The
509: classes of states with different quantum numbers of $J$ and $M$
510: are not mixed, and one
511: can study the onset of chaos, spectral characteristics, complexity
512: of wave functions and so on for each class separately. The
513: situation is similar, for example, to the Sinai billiard, where
514: the studies responsible for a hypothesis \cite{boh} of the
515: correspondence between quantum level statistics and classical
516: chaos can be tested using one octant of the billiard and
517: continuing wave functions to the entire area according to the
518: symmetry class. Combining states of various classes into a common
519: spectrum, one comes to the Poissonian level statistics
520: \cite{GuP,Brody}. To the best of our knowledge, correlations
521: between the states of different exact symmetry in the same
522: billiard were not studied.
523:
524: At the same time, serious efforts were applied to the problems of
525: approximate symmetries, onset of chaos along with destruction of
526: symmetry, transition from the GOE to GUE due to violation of
527: time-reversal invariance by the magnetic field or $T$-odd nuclear
528: forces, intermediate spectral statistics and so on \cite{Guhr}.
529: The example most relevant to nuclear structure is given by the
530: {\sl isospin invariance} \cite{HRM}. The classes of nuclear states
531: with different isospin are mixed by electromagnetic interactions
532: and strong forces violating charge symmetry, and this can be seen
533: in transition probabilities and reaction amplitudes. In the shell
534: model versions with exact isospin conservation and without weak
535: interactions, the classes of states are characterized by exact
536: quantum numbers $J^{\pi}T$.
537:
538: The common Hamiltonian that governs nuclear dynamics certainly
539: establishes correlations between the states of different classes
540: even if they belong to the region of quantum chaos. One can
541: imagine, for instance, a deformed system with extremely chaotic
542: many-body dynamics inside. Nevertheless, the rotational invariance
543: guarantees the existence of the rotational branch of the
544: excitation spectrum with energy $E_{J}$ at least approximately
545: given by $AJ(J+1)$. In a macroscopic system, this would be a
546: continuous Goldstone mode that emerges because of the spontaneous
547: orientational symmetry breaking by the choice of the body-fixed
548: frame; in a finite system it is simply a rotational band, or, for
549: a chaotic system, ``compound band" as suggested by Mottelson
550: \cite{Doss}. Thus, we obtain a clear correlation between the
551: states of the same band in different $J$-classes. The intrinsic
552: structures of these states should also be close. In a sense, this
553: might even be a classical rigid body with microscopic quantum
554: chaos of interacting constituents.
555:
556: \subsection{Geometric chaoticity}
557:
558: Consider a finite many-body system with exact angular momentum
559: conservation. The total spin of the system,
560: \begin{equation}
561: {\bf J}=\sum_{a} {\bf j}_{a}, \label{3.2}
562: \end{equation}
563: is built up of spins of individual constituents. As a number of
564: particles grows, so does the number of independent ways of
565: building a many-body state of a given total spin $J$. This number
566: determines the dimension of a given $J$-class, $d_{J}$; a similar
567: construction is necessary for a total isospin $T$. Those
568: independent combinations correspond to various recouplings of
569: spins in the process of constructing the full state:
570: \begin{equation}
571: \{[(j_{1}j_{2})j_{12}j_{3}]j_{123}j_{4}\}j_{1234}...J.
572: \label{3.3}
573: \end{equation}
574: Different paths to the same values of $J$ can be distinguished by
575: high $nj$-symbols or coefficients of fractional parentage.
576:
577: In the shell model, even with a particle number of the order of
578: 10, the number of different paths, eq. (\ref{3.3}), is large, and
579: the resulting products of many consecutive Clebsch-Gordan
580: coefficients (CGC) determine the orthogonal combinations within
581: the same class. Let us, for example, look at the class $J=0$ in a
582: system of an even particle number. We can start with a simple
583: state of seniority $s=0$, when we couple particles pairwise to
584: $j_{12}=j_{34}=...0$ (to guarantee the full permutational symmetry
585: it is more convenient to use the pair operators in secondary
586: quantization, as we do below). Here all CGC are trivial, and the
587: state has a very regular structure. Each new state should be
588: orthogonal to all previous ones, so that at some point we have to
589: employ another combination (a closed loop or few loops of vectors
590: ${\bf j}_{a}$) that proceeds through different intermediate
591: stages. At a sufficiently large dimension, the majority of paths
592: look as a {\sl random walk process of vector coupling}. This
593: source of randomness we call geometrical chaoticity.
594:
595: The property of geometrical chaoticity was practically used long
596: ago by Bethe \cite{Bethe}, see also \cite{Eric}, to derive the
597: partial nuclear level density $\rho(E;J)$ for a given angular
598: momentum in the model of noninteracting fermion gas. Assuming that
599: the projections $j_{z}=m$ of particle spins are coupled into the
600: total projection $M$ in a random walk process, one can apply the
601: central limit theorem and come to the Gaussian probability $w(M)$
602: of a given value of $M$ with zero mean and the variance
603: \begin{equation}
604: (\Delta M)^{2}=N\langle m^{2}\rangle, \label{3.4}
605: \end{equation}
606: expressed in terms of the (energy-dependent) number $N$ of active
607: fermions (particles and holes) and average single-particle value
608: of $m^{2}$ in space of available orbitals. The level density with
609: given $M$ is then given by $\rho(E;M)=\rho(E)w(M)$, where
610: $\rho(E)$ is the total level density. Using the standard trick,
611: one can get the approximate expression for the multiplicity
612: $d_{J}$ of states with given spin $J$,
613: \begin{equation}
614: \frac{d_{J}}{d}=w(J)-w(J+1)\approx \frac{2J+1}{2\sqrt{2\pi}
615: (\Delta M)^{3}}e^{-J(J+1)/2(\Delta M)^{2}}. \label{3.5}
616: \end{equation}
617: This result, invalid for the largest values of $J$, shows that the
618: maximum of the multiplicity is near $J=\Delta M-1/2$. From eq.
619: (\ref{3.4}) we see that it grows $\propto \sqrt{N}$.
620:
621: Regrettably, it is very hard to develop a statistical theory for a
622: random process with quantized vectors as ${\bf j}_{a}$ that are
623: coupled not algebraically. This would be equivalent to developing
624: statistical theory of fractional parentage coefficients. In spite
625: of various attempts and some useful results that can be extracted
626: from works by Wigner \cite{Wig}, Ponzano and Regge \cite{regge}
627: and others \cite{Bieden,Wong}, such theory still does not exist.
628:
629: The reality of geometric chaoticity is clearly seen in the nuclear
630: shell model \cite{big}. Prior to any diagonalization, one can
631: derive important characteristics of the energy spectrum directly
632: from the Hamiltonian matrix. A basis state $|JT;k\rangle$ of
633: independent particles {\sl projected} onto certain values of spin
634: and isospin is a superposition of the stationary states
635: $|JT;\alpha\rangle$ with the same real amplitudes $C^{\alpha}_{k}$
636: that determined the composition of the eigenstate in eq.
637: (\ref{2.2}). The weights $w^{\alpha}_{k}$ define the {\sl strength
638: function} $F_{k}(E)$ of the simple state $|JT;k\rangle$ according
639: to
640: \begin{equation}
641: F_{k}(E)=\sum_{\alpha}w^{\alpha}_{k}\delta (E-E_{\alpha}),
642: \label{3.6}
643: \end{equation}
644: where $E_{\alpha}$ are energies of the eigenstates. The strength
645: function (since $\sum_{k}F_{k}(E)=\rho(E)$, it is called {\sl
646: local density of states} in condensed matter theory, where the
647: basis states $|k\rangle$ are localized ones) determines the time
648: evolution of the state $|k\rangle$ prepared at the initial moment.
649: The centroid of the strength function $F_{k}(E)$ is given by the
650: {\sl diagonal} matrix element of the Hamiltonian, $H_{kk}$. The
651: energy dispersion $\sigma_{k}$ of the state $|JT;k\rangle$ can be
652: found as
653: \begin{equation}
654: \sigma^2_{k}=\langle k|(H-H_{kk})^{2}|k\rangle=\sum_{l\neq k}
655: H_{kl}^{2}, \label{3.7}
656: \end{equation}
657: the sum of all {\sl off-diagonal} matrix elements in the $k^{{\rm
658: th}}$ row of the Hamiltonian matrix. A remarkable fact is that in
659: a given shell model class $(JT)$ the dispersions $\sigma_{k}$ are
660: nearly constant, $\sigma_{k}\approx \bar{\sigma}$, for all states
661: $|JT;k\rangle$. This equilibration for {\sl noninteracting}
662: particles comes only from the $JT$-projection, and therefore is
663: the direct output of geometric chaoticity that accompanied the
664: projection algorithm. Parenthetically we can mention that the
665: constant magnitude of the dispersion (\ref{3.7}) is important for
666: determination of the {\sl spreading width} of the strength
667: function \cite{Fraz} that approaches the limit of $2\bar{\sigma}$
668: in the case of strong fragmentation.
669:
670: \subsection{Rotationally invariant two-body Hamiltonian}
671:
672: Now we explicitly introduce the requirements of rotational
673: invariance in the Hamiltonian (\ref{2.1}), both for
674: single-particle states and the interaction. We assume spherical
675: symmetry of the mean field with orbitals characterized by the
676: angular momentum $j$, its projection $j_{z}=m$, and isospin
677: projection $\tau_{3}$. For definitiveness we consider fermions
678: with isospin 1/2 and assume that every value of $j$ appears only
679: once. The Hamiltonian of the system is determined by the set of
680: single-particle energies $\epsilon_{j}$ that, under conditions of
681: rotational and isospin invariance, do not depend on $m$ and
682: $\tau_{3}$, and by the interaction that preserves the total
683: angular momentum $L$ and total isospin $t$ of the interacting
684: pair. The eigenstates of the Hamiltonian have exact quantum
685: numbers of total angular momentum, $J$ and $J_{z}=M$, and total
686: isospin, $T$ and $T_{3}$.
687:
688: The most general form of the two-body interaction under such
689: conditions is
690: \begin{equation}
691: H_{int}=\sum_{L\Lambda,tt_{3};\{j\}}V_{Lt}(j_{1}j_{2};j_{3}j_{4})
692: P^{\dagger}_{L\Lambda,tt_{3}}(j_{1}j_{2})P_{L\Lambda,tt_{3}}(j_{3}j_{4}).
693: \label{3.8}
694: \end{equation}
695: Here we use the pair annihilation and creation operators for each
696: $Lt$-pair channel,
697: \[P_{L\Lambda,tt_{3}}(j_{1}j_{2})=\frac{1}{\sqrt{1+\delta_{j_{1}j_{2}}}}
698: [a_{j_{1}}a_{j_{2}}]_{L\Lambda,tt_{3}},\]
699: \begin{equation}
700: P^{\dagger}_{L\Lambda,tt_{3}}(j_{1}j_{2})=\frac{1}
701: {\sqrt{1+\delta_{j_{1}j_{2}}}}
702: [a^{\dagger}_{j_{2}}a^{\dagger}_{j_{1}}]_{L\Lambda,tt_{3}},
703: \label{3.9}
704: \end{equation}
705: where the vector coupling with the appropriate CGC to the total
706: rotational ($L\Lambda$) and isospin ($tt_{3}$) quantum numbers of
707: the pair is implied, Fig. \ref{diagram}. One can also remove isospin
708: invariance taking different values for proton and neutron
709: single-particle levels and interaction matrix elements but still
710: preserving angular momentum conservation.
711:
712: \begin{figure}
713: %FIG 1
714: \begin{center}
715: %\vskip 0.5 cm
716: \includegraphics[width=7 cm]{diagram.eps}
717: \end{center}
718: \caption{\label{diagram}The diagram of the two-body interaction in
719: particle-particle and particle-hole channels characterized by the
720: total spins, $L$ and $K$, respectively.}
721: \end{figure}
722:
723: In this formulation, the numerical parameters of the Hamiltonian
724: are, apart from the single-particle energies, the interaction
725: amplitudes $V_{Lt}$ that do not depend on projections $\Lambda$
726: and $t_{3}$. The number $k$ of these amplitudes is typically much
727: smaller than the number $d_{JT}$ of levels in the $JT$-class (the
728: classes with the largest possible values of $J$ and $T$ might be
729: exceptional in this respect). Thus, the well studied nuclear $sd$
730: shell model includes $d_{5/2}, s_{1/2}$ and $d_{3/2}$ orbitals for
731: neutrons and protons. This model is completely defined by three
732: single-particle energies and 63 interaction matrix elements
733: $V_{Lt}$ while, for example, with 12 valence fermions, there are
734: 839 states in the class with lowest quantum numbers $JT=00$
735: \cite{big}, the largest class in this model appears with $JT=31$
736: and has $d_{JT}=6706$.
737:
738: It is always possible to recouple the creation and annihilation
739: operators from the particle-particle channel used in eqs.
740: (\ref{2.1}) and (\ref{3.8}) to the particle-hole channel, see Fig.
741: \ref{diagram}, where the Hamiltonian would have the multipole-multipole
742: structure $\{[a^{\dagger}a]_{K} [a^{\dagger}a]_{K}\}_{00}$ with
743: the spin $K$ of the particle-hole pairs; the similar recoupling is
744: performed for isospins. The corrections to the orbital energies
745: (one-body terms) appear in this process. The transformation can be
746: done in two ways which correspond to the crossing transition from
747: the $s$-channel to $t$- and $u$-channels in quantum field theory.
748: It is important to stress that, although not seen immediately, the
749: number of parameters is the same in different channels because of
750: the permutational symmetry of identical particles, as will be
751: shown better in the next subsection.
752:
753: The typical class dimensions of few hundred up to few thousand
754: make the nuclear shell model a perfect tool for studying the exact
755: solution of a many-body problem. With dimensions of this size, the
756: diagonalization is straightforward and we obtain all individual
757: eigenfunctions while the statistical regularities are already
758: clearly pronounced. Thus, we are in a typical realm of mesoscopic
759: physics. Again we stress that {\sl all} classes of states are
760: described by the same matrix elements and therefore have to be
761: correlated.
762:
763: The particle-particle interaction channel with $L=0$ and $t=1$,
764: when $j_{1}=j_{2}$ and $j_{3}=j_{4}$, represents conventional
765: isospin-invariant isovector pairing. This component of the
766: interaction is, as a rule, considered to be responsible for the
767: main pairing effects in binding energy. It is possible that around
768: the $N=Z$ line the isoscalar pairing, $t=0$, is important,
769: especially in nuclei far from stability; for classification of
770: various types of pairing see, for example \cite{good} and
771: references therein. According to the generalized Pauli principle
772: for a two-nucleon system, the symmetries in spin-spatial variables
773: and in isospin are complementary. The symmetry relation for the
774: pair operators (\ref{3.9}) reads
775: \begin{equation}
776: P_{L\Lambda,tt_{3}}(j_{1}j_{2})=(-)^{j_{1}+j_{2}+L+t}
777: P_{L\Lambda,tt_{3}}(j_{2}j_{1}). \label{3.10}
778: \end{equation}
779: In particular, for $j_{1}=j_{2}$, only even values $L=0,2,...,
780: 2j-1$ are allowed for $t=1$ (or for identical fermions without
781: isospin) while $t=0$ requires the odd values $L=1,3,...,2j$. With
782: time reversal invariance, the parameters $V_{Lt}$ can be chosen
783: real.
784:
785: \subsection{Single $j$-level}
786:
787: Here we give the formalism for the simplest fermionic space,
788: namely that of $\Omega=2j+1$ single-particle states $|jm)$ of a
789: single $j$-level for one kind of particles. With this
790: simplification, it will be possible to see the core of the
791: problem. In this case the mean field part is just a constant
792: proportional to the particle number, and the Hamiltonian takes the
793: form
794: \begin{equation}
795: H=\epsilon N+\sum_{L}V_{L}\sum_{\Lambda}P^{\dagger}_{L\Lambda}
796: P_{L\Lambda}, \label{3.11}
797: \end{equation}
798: where the pair operators are defined in terms of the $3j$-symbols
799: as
800: \begin{equation}
801: P^{\dagger}_{L\Lambda}=\frac{1}{\sqrt{2}}\sum_{m_{1}m_{2}}
802: \sqrt{2L+1}(-)^{L-\Lambda}\left(\begin{array}{ccc}
803: j & L & j\\
804: m_{1} & -\Lambda & m_{2}\end{array}\right)
805: a^{\dagger}_{1}a^{\dagger}_{2}. \label{3.12}
806: \end{equation}
807: {\sl Only even} $L$ pairs are present in eq. (\ref{3.11}) so that
808: the number of independent interaction parameters $V_{L}$ is
809: $k=j+1/2$; all unnecessary labels are omitted.
810:
811: We define the multipole operators in the particle-hole channel,
812: Fig. \ref{diagram}, as
813: \begin{equation}
814: M_{K\kappa}=\sum_{m_{1}m_{2}}(-)^{j-m_{1}}\left(\begin{array}{ccc}
815: j & K & j\\
816: -m_{1} & \kappa & m_{2}\end{array}\right)a^{\dagger}_{2}a_{1}.
817: \label{3.13}
818: \end{equation}
819: Here any integer value of $K$ from 0 to $2j$ is allowed. The
820: Hermitian conjugation gives
821: \begin{equation}
822: M^{\dagger}_{K\kappa}=(-)^{\kappa}M_{K-\kappa}. \label{3.14}
823: \end{equation}
824: The special important cases are the particle number operator
825: \begin{equation}
826: N=\sum_{m}a^{\dagger}_{m}a_{m}=\sqrt{\Omega}\,M_{00},
827: \label{3.14a}
828: \end{equation}
829: and the angular momentum (in spherical components)
830: \begin{equation}
831: J_{\kappa}=\sum_{mm'}(m|j_{\kappa}|m')a^{\dagger}_{m}a_{m'}=
832: \sqrt{j(j+1)\Omega}\,M_{1\kappa}. \label{3.14b}
833: \end{equation}
834:
835: The alternative, particle-hole, form of the Hamiltonian is
836: \begin{equation}
837: H=\tilde{\epsilon}N-\frac{1}{2}\sum_{K}\tilde{V}_{K}\sum_{\kappa}
838: M^{\dagger}_{K\kappa}M_{K\kappa}, \label{3.15}
839: \end{equation}
840: where the single-particle energy is renormalized,
841: \begin{equation}
842: \tilde{\epsilon}=\epsilon +
843: \frac{1}{2\Omega}\sum_{K}\tilde{V}_{K}, \label{3.16}
844: \end{equation}
845: and the interaction parameters are transformed according to
846: \begin{equation}
847: \tilde{V}_{K}=(2K+1)\sum_{L}(2L+1)\left\{\begin{array}{ccc}
848: j & j & L\\
849: j & j & K\end{array}\right\}V_{L}. \label{3.17}
850: \end{equation}
851: Inversely,
852: \begin{equation}
853: V_{L}=\sum_{K}\left\{\begin{array}{ccc}
854: j & j & K\\
855: j & j & L\end{array}\right\}\tilde{V}_{K}. \label{3.18}
856: \end{equation}
857:
858: Since a reversible algebraic transformation cannot increase the
859: number of independent parameters, there should exist constraints
860: that reduce the number of independent constants $\tilde{V}_{K}$
861: from $2j+1$ to the number $k=j+1/2$ of the original parameters
862: $V_{L}$. Indeed, the particle-hole amplitudes $\tilde{V}_{K}$ are
863: {\sl interrelated} through
864: \begin{equation}
865: \tilde{V}_{K}=(2K+1)\sum_{K'}(-)^{K+K'}\left\{\begin{array}{ccc}
866: j & j & K\\
867: j & j & K'\end{array}\right\}\tilde{V}_{K'}. \label{3.19}
868: \end{equation}
869: These relations expressing the symmetry of recoupling between the
870: $t$- and $u$-channels are important for the dynamics of the model.
871: They were discussed more in detail by one of the authors
872: \cite{Volya}. In particular, we note the results for the pairing
873: interaction, $L=0$,
874: \begin{equation}
875: V_{0}=-\frac{1}{\Omega}\sum_{K}(-)^{K}\tilde{V}_{K}, \label{3.20}
876: \end{equation}
877: and monopole interaction, $K=0$,
878: \begin{equation}
879: \tilde{V}_{0}=-\frac{1}{\Omega}\sum_{K}\tilde{V}_{K}. \label{3.21}
880: \end{equation}
881:
882: To conclude this section, we write down the commutator algebra of
883: the pair and multipole operators (we use the abbreviation $g_{K}
884: =\sqrt{2K+1}$),
885: \begin{equation}
886: [P_{L'\Lambda'},P^{\dagger}_{L\Lambda}]=\delta_{LL'}
887: \delta_{\Lambda\Lambda'}+2\sum_{K\kappa}g_{K}^{2}
888: X^{LL';K}_{\Lambda \Lambda'\kappa}M^{\dagger}_{K\kappa};
889: \label{3.22}
890: \end{equation}
891: \begin{equation}
892: [P^{\dagger}_{L\Lambda},M_{K\kappa}]=2\sum_{L'\Lambda'}
893: X^{LL';K}_{\Lambda \Lambda'\kappa} P^{\dagger}_{L'\Lambda'};
894: \label{3.23}
895: \end{equation}
896: \begin{equation}
897: [M^{\dagger}_{K\kappa},M_{K'\kappa'}]=\sum_{S\sigma}[1-(-)^{K+K'+S}]
898: \frac{g_{S}^{2}}{\sqrt{g_{K}g_{K'}}}
899: X^{KK';S}_{\kappa\kappa'\sigma} M^{\dagger}_{S\sigma}.
900: \label{3.24}
901: \end{equation}
902: We introduced here the common geometric factor
903: \begin{equation}
904: X^{LL':K}_{\Lambda\Lambda'\kappa}=g_{L}g_{L'}
905: \left\{\begin{array}{ccc}
906: L & L'& K\\
907: j & j & j\end{array}\right\}(-)^{\Lambda}
908: \left(\begin{array}{ccc}
909: L & L' & K\\
910: -\Lambda & \Lambda' & \kappa\end{array}\right). \label{3.25}
911: \end{equation}
912: In the right hand side of eq. (\ref{3.22}), we did not indicate
913: explicitly the symmetry factors $\Theta_{L}=[1+(-)^{L}]/2$ and
914: $\Theta_{L'}$; similarly eq. (\ref{3.23}) contains $\Theta_{L}$.
915:
916: The closed algebra of particle-particle and particle-hole
917: operators is too complicated for a general analysis. However, it
918: contains closed subalgebras with simpler properties. The operators
919: $P_{00},\,P^{\dagger}_{00}$ and $M_{00}\propto N$ form the well
920: known from seniority theory \cite{Racah,Talmi} {\sl quasispin
921: algebra} isomorphic to $SU(2)$ and widely used for the solution of
922: the pairing problem, approximate in BCS theory \cite{Bel} or exact
923: \cite{EP}. The odd-$K$ multipoles $M_{K\kappa}$ form the algebra
924: $U(2j+1)$, and the three components of angular momentum
925: proportional to $M_{1\kappa}$ give a standard $SU(2)$. The
926: algebraic properties of the operators will be used in the
927: equations of motion.
928:
929: \section{Ordered spectra from random interactions?}
930:
931: \subsection{Main evidence}
932:
933: As mentioned in Introduction, the shell model calculations with
934: random two-body interactions (\ref{3.8}) provide unexpected
935: results. Using the $sd$- and $pf$-shell model for even-even nuclei
936: with various particle numbers and an ensemble of random parameters
937: $V_{Lt}$, the authors of Ref. \cite{JBD} found a surprisingly
938: large fraction of cases with the ground state spin $J_{0}=0$.
939: Being later confirmed and studied by many authors for different
940: ensembles and different fermion and boson single-particle spaces,
941: the effect seems to be generic, see for example
942: \cite{JBDT,BFP,BF,Bij,KZC,Kus,KSJ,Mul,Hor,oddA,Mul1,Cov,Yad,Mul2,Zhao,Zhao1,Zhao2,Arima,Yoshi,Zhao3,Droz,vela,Zhaocomm,velah,KAPP,KapPJ,Chau}.
943:
944: The naive idea of what should be the ground state spin of a system
945: governed by random rotationally invariant interactions comes from
946: a simple counting of multiplicities $d_{J}(N)$ of states with a
947: given spin $J$ and particle number $N$ in Hilbert many-body space
948: spanned by a given set of single-particle orbitals. As stated in
949: Subsection 3.2, as a result of a random spin coupling, the spin
950: value with the largest multiplicity increases with the particle
951: number $\propto\sqrt{N}$. However, the empirical probabilities
952: $f_{J}$ of the ground state spin $J$ turn out to be very different
953: from a simple estimate $d_{J}/d$, where $d$ is the total dimension
954: of space. The original paper \cite{JBD} gives the following
955: results of direct repeated diagonalization and averaging over the
956: ensemble (we discuss the choice of the ensemble later on): for
957: $N=6$ identical particles in the $sd$-shell ($\Omega=12$) they
958: found $f_{0}=76\%$ and for the $pf$-shell ($\Omega=20$)
959: $f_{0}=75\%$, whereas the corresponding Hilbert space
960: multiplicities are equal to $d_{0}/d=9.8\%$ and 3.5\%,
961: respectively. Similar results were found for two kinds of
962: particles, the state with $J_{0}=T_{0}=0$ had a predominant
963: probability to be found as the ground state.
964:
965: \begin{table}
966: $$
967: \begin{array}{|r||r|r|r|r|r|r|r|r|r|}
968: \hline
969: J& (a) & (b) & (c) & (d) & (e) & (f) & (g) &(h)&(i) \\
970: \hline
971: 0&0.61&12.7 & 65.4 & 61.9 & 65.3 & 54.5 & 80.5 & 55.2 & 64.1 \\
972: 2&1.45&3.6 & & 0.8 & 0.8 & & & 0.6 & \\
973: 4&2.38&5.4 & 1.9 & 2.5 & 2.7 & & 1.0 & 3.7 & 2.2 \\
974: 5&2.15&1.8 & & & & 9.1 & & & \\
975: 6&3.18&6.4 & 4.8 & 6.5 & 14.7 & 9.1 & 1.7 & 6.5 & 4.9 \\
976: 8&3.74&3.6 & 3.4 & 2.6 & 2.2 & & 1.8 & 4.7 & 3.3 \\
977: 10&4.41&4.1 & 2.6 & 3.0 & 5.8 & 9.1 & 1.2 & 3.5 & 2.4 \\
978: 12&4.53&4.4 & & & 1.3 & & & 1.0 & \\
979: 13&4.07&2.6 & & & & & & 0.6 & 0.6 \\
980: 16&4.49&3.0 & & & 0.7 & & & 0.7 & \\
981: 18&4.31&3.1 & 1.0 & 1.7 & & & & 1.3 & 1.0 \\
982: 28&2.05&1.4 & 0.9 & 1.2 & & 9.1 & & 0.9 & 1.0 \\
983: 33&0.94&0.9 & & & & & & 0.6 & \\
984: 36&0.66&0.9 & & 0.7 & & & & & \\
985: 42&0.19&0.5 & 0.8 & 0.6 & & & & 0.9 & 0.7 \\
986: 46&0.05&0.3 & 0.8 & 1.0 & & & & 0.8 & 0.7 \\
987: 48&0.05&0.3 & 11.8& 11.5 & 1.4 & 9.1 & 9.2 & 12.4 & 11.7 \\
988: \hline
989: \end{array}
990: $$
991: %TABLE 1.
992: \caption{\label{tab1}Statistics of ground state spins $J_{0}$ for
993: $N=6$ identical particles on a single $j=21/2$ orbital: $(a)$
994: multiplicity $d_{J}/d$; (b) predictions from approximating fermion
995: pairs with non-interacting bosons; $(c)$ fractions $f_{J}$ for a
996: random ensemble with the uniform distribution of all $V_{L}$ in
997: the interval [-1,1]; $(d)$ for a Gaussian random ensemble of
998: $V_{L}$ with zero mean and dispersion equal to 1; $(e)$ for the
999: uniform distribution of $V_{L}$ scaled by $(2L+1)^{-1}$; $(f)$
1000: predictions according to a recipe of Ref. \cite{Zhao2}; $(g)$ for
1001: the uniform ensemble of $V_{L\neq 0}$ and fixed attractive
1002: pairing, $V_{0}=-1$; $(h)$ the same as in $(g)$ but with repulsive
1003: pairing, $V_{0}=+1$; $(i)$ the same as in $(g)$ and $(h)$ but
1004: without pairing, $V_{0}=0$. Except for columns $(a)$ and $(b)$,
1005: only fractions $f_{J}>0.5\%$ are included. }
1006: \end{table}
1007:
1008: To display the universality of the effect we show in Table
1009: \ref{tab1} the results of the diagonalization for various random
1010: ensembles for a system of $N=6$ identical fermions on a single
1011: level $j=21/2$. Comparing the column $(a)$ with $(c)$ and $(d)$, we see that
1012: the approximately the same predominance of $J_{0}=0$ (it exceeds
1013: the statistical multiplicity by an order of magnitude) occurs for
1014: the uniform and Gaussian ensembles of the parameters $V_{L}$.
1015: Moreover, even the suppression of high-$L$ components of the
1016: interaction by a factor $(2L+1)^{-1}$ does not change $f_{0}$,
1017: column $(e)$. Contrary to the statement of Ref. \cite{JBD} that
1018: the choice of ensemble is crucial, we come to the conclusion
1019: confirmed by other works that the predominance of $J_{0}=0$ is
1020: insensitive to the specific features of the random ensemble.
1021:
1022: The Gaussian ensemble used in \cite{JBD} assumed that the
1023: interaction parameters are normally distributed uncorrelated
1024: random variables with zero mean and the variance that has an extra
1025: factor of 2 for the diagonal matrix elements. The latter property
1026: was borrowed from canonical Gaussian ensembles and in reality does
1027: not matter. An assumption was also made concerning the choice of
1028: the variance $\overline{V_{L}^{2}}$ scaled as a function of $L$ as
1029: $(2L+1)^{-1}$. According to the original idea of Ref. \cite{JBD},
1030: this choice was made in order to obtain the most random
1031: interaction with analogous statistical properties in the
1032: particle-particle and particle-hole channels (``Random
1033: Quasiparticle Ensemble", RQE). Regrettably, this idea cannot be
1034: implemented since such a scaling is not invariant under the
1035: transformation (\ref{3.17}) and (\ref{3.18}). The reason is in a
1036: formally different number of parameters in the two channels that
1037: is equalized by extra constraints on the parameters in the
1038: particle-hole channel that are not independent, eq. (\ref{3.19}).
1039: The non-equivalence of the channels could be easily seen if the
1040: definition for the variance would be written in the way explicitly
1041: including the factor $[1+(-)^{L}]/2$ necessary for Fermi
1042: statistics, eq. (\ref{3.10}). Therefore, the RQE is just one of
1043: possible choices without preferential meaning.
1044:
1045: \begin{figure}
1046: \begin{center}
1047: %FIG 2
1048: %\vskip 0.5 cm
1049: \includegraphics[width=14 cm]{pstat.eps}
1050: \end{center}
1051: \caption{Fractions of states with ground state spins $J_{0}=0$
1052: (circles connected by solid lines) and $J_{0}=J_{{\rm max}}$
1053: (squares connected by dashed lines) for the single$-j$ models as a
1054: function of $j$. The three panels correspond to particle numbers
1055: $N=4,6$ and 8, from left to right, respectively. The uniform
1056: distribution of interaction parameters $V_{L}\in [-1,1]$ was used.
1057: \label{pstat}}
1058: \end{figure}
1059:
1060: %[Yours Fig. 2 - only panels a,b,c,d]
1061: %FIG 3
1062: \begin{figure}
1063: \begin{center}
1064: \includegraphics[width=11 cm]{N4N5.eps}
1065: \end{center}
1066: \caption{\label{N4N5}Distributions $f_{J}$ of ground state spins:
1067: $(a)$ 4 fermions on the $j=15/2$ level; $(b)$ 5 fermions on the
1068: $j=15/2$ level; in this odd-$N$ system, in accordance to panel
1069: $(a)$, the maximum of the probability corresponds to $J=j$; $(c)$
1070: 6 fermions on the $j=11/2$ level; $(d)$ 5 bosons with interaction
1071: in the pair states with $L=0,2,4,6,8,10$, see Section 5. The
1072: uniform distribution of $V_{L}$ was used. The dotted lines in
1073: panels $(a),(b)$ and $(d)$ indicate the statistical distribution
1074: of multiplicities $d_{J}/d$; the dashed line in panel $(c)$
1075: corresponds to the case with no pairing, $V_{0}=0$.}
1076: \end{figure}
1077:
1078: Various assumptions concerning the choice of ensemble lead to
1079: quite similar results for $f_{J}$ although details can differ.
1080: Along with this, it was observed \cite{BF,Mul} that in many cases
1081: the fraction of the ground state spins equal to the {\sl largest
1082: possible|} spin $J_{{\rm max}}$ is also considerably enhanced, see
1083: the last line of Table \ref{tab1}. Note that in the single-$j$
1084: model (the results for $f_{0}$ and $f_{J_{{\rm max}}}$ in this
1085: case are shown in Fig. \ref{pstat} as a fluctuating function of
1086: $j$) the state with $J=J_{{\rm max}}$ is {\sl unique} being
1087: constructed by full alignment of the particles along the
1088: quantization axis (no random walk in this case). The effects
1089: persist for odd-$A$ \cite{Mul}, see also Fig. \ref{N4N5}$b$, and
1090: odd-odd \cite{oddA} nucleonic systems and for interacting bosons
1091: \cite{BF}, see also Fig. \ref{N6N7}. Typical examples are shown in
1092: Fig. \ref{N4N5}. Structures of the wave functions, properties of
1093: the observables and excitation spectra also were studied in a
1094: multitude of models. We will discuss the most important findings
1095: as we go step by step testing the explanations put forward by
1096: various authors.
1097:
1098: \subsection{Induced pairing?}
1099:
1100: The first idea suggested for the explanation of the ground state
1101: spin effects \cite{JBD,JBDT} was related to Cooper-type pairing
1102: induced by random interactions through some high-order mechanism.
1103: For a single $j$-level there are deep symmetry reasons to presume
1104: a special role of pairing since the seniority quantum number is
1105: only broken by about $1/3$ of $k=j+1/2$ linearly independent
1106: combinations of interaction parameters \cite{Volya,rowe}. The
1107: pairing idea was supported \cite{JBD} by the enhanced gap between
1108: the ground and first excited states in the case of $J_{0}=0$,
1109: presence of odd-even staggering and the large pair transfer matrix
1110: element $\langle N-2|P|n\rangle$ between the ground states of
1111: adjacent even nuclei. The pair transfer operator $P$, however, was
1112: taken in such a form (different in different realizations) that in
1113: fact predetermined the result.
1114:
1115: Table \ref{tab1} contains important information about the role of
1116: explicit pairing (parameter $V_{0}$ in the Hamiltonian) in the
1117: single-$j$ case. Eliminating pairing completely from the dynamics,
1118: column $(i)$, does not noticeably change the fraction $f_{0}$.
1119: Even the transition to ``antipairing" (fixed $V_{0}=+1$) only
1120: slightly reduces the value of $f_{0}$, column $(h)$. The
1121: attractive fixed pairing, $V_{0}=-1$, however, increases the
1122: fraction $f_{0}$ to 80\%, column $(g)$. Similar results can be
1123: seen in Fig. \ref{N4N5}$c$: in this case, $N=6$ and $j=11/2$,
1124: fully random interactions lead to $f_0=55.9$\%, setting pairing to
1125: zero $V_0=0$ reduces fraction of $J_0=0$ ground states to
1126: $f_0=51.9$\%, in ``antipairing'' case $f_0=44.4$\% and finally,
1127: forced pairing with $V_0=-1$ leads to $f_0=82.7$\%.
1128:
1129: The fact that, as a rule, the ground state with $J_{0}=0$ does not
1130: contain considerable pairing correlations, can be seen from the
1131: observation based on the ICE, Section 2.4. Indeed, with forced
1132: pairing (by means of large negative $V_0$), many random
1133: realizations change the ground state structure undergoing a
1134: transition to the superconducting paired state. The ICE can be
1135: used here to study such transitions in each individual
1136: realization. In Fig. \ref{miscice}, left, the behavior of ICE for a few
1137: randomly selected realizations is shown as a function of the
1138: pairing strength $V_0$, around which the fluctuations necessary to
1139: obtain the ICE are imposed. All selected realizations had $J_0=0$
1140: even without pairing, i.e. at $V_0=0.$ The peaks are observed in
1141: the ICE curves for all random realizations. The location of the
1142: peak roughly indicates the critical point of the phase transition
1143: with the maximum sensitivity to noise, while the sharpness of the
1144: peak and ICE magnitude reflect the size of the critical region. As
1145: one would expect, the properties of the pairing phase transition
1146: vary significantly from one sample to another. It is however clear
1147: that, in order for a significant fraction of random realizations
1148: to exhibit developed pairing, an average coherent attraction in
1149: the $V_0$ channel must be added, see right
1150: panel in Fig. \ref{miscice}, an analog of a {\sl displaced}
1151: ensemble \cite{vela,velah}.
1152:
1153: \begin{figure}
1154: \begin{center}
1155: %\vskip 0.5 cm
1156: \includegraphics[width=14 cm]{miscice.eps}
1157: \end{center}
1158: \caption{\label{miscice} On the left panel the invariant correlational
1159: entropy (ICE)
1160: for randomly selected realizations of ensemble is plotted
1161: as a function of pairing strength $V_0$. The system of six
1162: particles on $j=15/2$ level is chosen. Right panel shows the distribution of
1163: critical values $V_0$ defined as peaks on each ICE curve.}
1164: \end{figure}
1165:
1166: To find out if there is an important role of induced pairing, we
1167: compare directly the empirical ground state wave functions
1168: $|0\rangle$ with $J_{0}=0$ from the diagonalization in the random
1169: ensemble to the fully paired state $|s=0\rangle$ of seniority zero
1170: and the same particle number that can be built uniquely for a
1171: single-$j$ level. Fig. \ref{rpairing} shows with shaded histograms
1172: the distribution $P(x)$ of the
1173: overlaps
1174: \begin{equation}
1175: x=|\langle 0|s=0\rangle|^{2}, \quad 0\leq x\leq 1, \label{4.1}
1176: \end{equation}
1177: for 4 and 6 fermions on the $j=15/2$ level, left and right panels,
1178: respectively, that have $d_{0}=3$ and $d_{0}=4$ states of $J=0$.
1179: The overlap (\ref{4.1}) is one of the weights $w$ of the wave
1180: function, namely the one for the paired basis state. The paired
1181: structure would give a peak of $P(x)$ at $x\rightarrow 1$ while a
1182: random wave function is characterized by the distribution $P(x)$
1183: of eq. (\ref{2.6}). For $d=3$, left panel,
1184: \begin{equation}
1185: P_{d=3}(x)=\frac{1}{2\sqrt{x}}, \label{4.2}
1186: \end{equation}
1187: with a peak at $x\rightarrow 0$ shown with solid line. This
1188: distribution appears in the problem of pion multiplicity from a
1189: disordered chiral condensate \cite{dcs,ocs}, where isospin 1
1190: determines the dimension $d=3$. Similarly, in the right panel,
1191: with a bigger space, $d_{0}=4$, the chaotic distribution is
1192: $P_{d=4}(x)\propto (1-x)^{3/2}x^{-1/2}$, and the empirical
1193: distribution displays only a slight excess near $x=1$, of the
1194: order of 1\% in the total normalization.
1195:
1196: %Fig.5
1197: \begin{figure}
1198: \begin{center}
1199: %\vskip 0.5 cm
1200: \includegraphics[width=14 cm]{rpairing.eps}
1201: \end{center}
1202: \caption{\label{rpairing} The distributions of overlaps $x$
1203: between the ground states of spin zero in the random ensemble and
1204: states of seniority zero, for $N=4$, left, and $N=6$, right,
1205: fermions on the $j=15/2$ level, shaded histograms, for the
1206: ensemble of all random $V_{L}$. The solid lines show the
1207: statistical distributions expected for a chaotic system of
1208: dimensions $d=3$, left, and $d=4$, right. The unshaded
1209: distributions are calculated for the parts of the ensemble that
1210: give a ground state $J=0$ {\sl and} the first excited state
1211: $J=2$.}
1212: \end{figure}
1213:
1214: A similar analysis with similar results was also carried out for
1215: the $sd$ shell model space \cite{Hor}, where the geometry is much
1216: richer and includes also isospin. The calculation of information
1217: entropy (\ref{2.3}) revealed the chaotic character of eigenstates;
1218: even the ground state always has a high degree of complexity
1219: exceeding that for the realistic shell model. One can compare the
1220: ground state wave functions $|JT=00\rangle$ obtained with random
1221: interactions to those corresponding to the realistic system. With
1222: the standard effective interaction for this shell \cite{BW} (and
1223: the results were shown to change only slightly for different
1224: interactions), the comparison for $^{24}$Mg ($N=8$ nucleons)
1225: provides the following average overlaps $x$ with the ground state
1226: wave functions for different random ensembles: (a) degenerate
1227: single-particle energies and all 63 two-body matrix elements
1228: generated as random uniformly distributed quantities:
1229: $f_{0}=0.591,\, x=0.020$; (b) realistic single-particle energies
1230: and random interaction matrix elements: $f_{0}=0.493,\,x=0.053$;
1231: (c) realistic single-particle energies and six pairing interaction
1232: matrix elements with $L=0$ and $t=1$, and random remaining 57
1233: matrix elements: $f_{0}=0.678,\,x=0.106$; (d) single-particle
1234: energies and 57 non-pairing matrix elements set to zero, while the
1235: pairing matrix elements taken as random variables:
1236: $f_{0}=0.922,\,x=0.052$. In the cases of realistic single-particle
1237: levels and random interactions, the ensembles of two-body
1238: amplitudes were chosen in such a way that have a realistic ratio
1239: of their magnitude to the single-particle level spacings
1240: \cite{Hor} in order to allow for fair comparison.
1241:
1242: Although the large value of $f_{0}$ is common for all variants, we
1243: see again that the presence of regular pairing, case (c),
1244: increases the fraction $f_{0}$. The largest $f_{0}$ is observed in
1245: case (d), where the off-diagonal pair transfer amplitudes make
1246: quantum numbers $J=T=0$ preferable for an even number of pairs,
1247: whereas the competing influence of incoherent interactions and the
1248: mean field level splitting is absent. The average overlaps $x$ are
1249: small in confirmation of the conclusion carried over from the
1250: single-$j$ model that the ground state wave functions with random
1251: interactions are far away from realistic ones which are up to high
1252: extent determined by pairing, although presence of deformation in
1253: realistic nuclei is another factor reducing average $x$.
1254:
1255: In systems with many double degenerate orbitals for spins $j=1/2$
1256: \cite{KapPJ}, the fraction $f_{0}$ turns out to be close to 100\%,
1257: which, at least partly, is, like in the previous case (d), induced
1258: by a great preponderance of off-diagonal pair transfers. Here one
1259: should mention that such a system reminds a quantum spin glass
1260: with random spin-spin interactions. In that case \cite{Sush} the
1261: ground state spin increases $\propto\sqrt{N}$ as expected for
1262: random spin coupling. The crucial difference as compared to
1263: shell-model systems is in the type of random coupling. For quantum
1264: spin glasses the spin-spin interaction $({\bf s}_{1}\cdot{\bf
1265: s}_{2})$ is usually assumed. This is equivalent to a fixed
1266: relation 3:1 between the singlet and triplet parts of the
1267: interaction. Contrary to that, in shell-model systems those parts
1268: are fully uncorrelated.
1269:
1270: One can notice that even in case (a) the average overlap of 2\% is
1271: higher than what we would expect, eq. (2.6), from the uniform
1272: distribution of the components, $\sim 1/d\sim 0.3\%$ for the
1273: actual dimension $d=325$. The maximum effect of 11\% is reached in
1274: case (c) due to the combination of two effects. First, the
1275: presence of realistic pairing lowers energies of states with
1276: paired particles. Second, the effective dimension $d^{\ast}$ is
1277: now smaller than $d$ because the contributions of non-paired
1278: states to the ground state are appreciably reduced. The
1279: stabilizing presence of the mean-field orbitals, case (b), also
1280: increases the overlap with the shell model ground state. To
1281: conclude, a small effect of induced pairing should be present but
1282: not as a main reason for the statistical predominance of $f_{0}$.
1283:
1284: \subsection{Time-reversal invariance?}
1285:
1286: The normal pairing ($Lt=01$ channel) is believed to be singled out
1287: as the most important part of residual nucleon-nucleon forces due
1288: to the maximum overlap of spatial wave functions of the paired
1289: particles \cite{BMP}. The coherent effects of this residual
1290: attraction are enhanced by a greater density of states for such
1291: pairs since in this case any single-particle state $|1)$ is
1292: coupled to its time-reversed counterpart $|\tilde{1})$ that has,
1293: in the absence of external magnetic or Coriolis fields, exactly
1294: the same single-particle energy (Kramers degeneracy). The same
1295: reasoning is behind the assigning a special role to the Cooper
1296: pairs with zero total momentum and singlet spin state in
1297: superconductors.
1298:
1299: Since the dynamical specificity of pairing forces converts the
1300: ground state into a condensate of time-reversed pairs, it is
1301: natural to invert the problem and hypothesize that the
1302: time-reversal invariance can imply the preponderance, at least in
1303: the statistical sense, of the ground states with $J_{0}=0$
1304: \cite{vela}. This idea was checked by Bijker, Frank and Pittel
1305: \cite{BFP} who explicitly included in the random two-body
1306: interaction a Hermitian but not $T$-invariant part assuming the
1307: interaction Hamiltonian in the form that was earlier studied as an
1308: example of the transition from the GOE to the GUE,
1309: \begin{equation}
1310: H=\cos\alpha\,H^{R}+i\sin\alpha\,H^{I}, \label{4.4}
1311: \end{equation}
1312: where $H^{R}$ and $H^{I}$ are uncorrelated Gaussian variables with
1313: zero mean and the same variance of off-diagonal elements. Here,
1314: because of Hermiticity, the matrix elements of $H^{R}$ should be
1315: real and symmetric, and those of $H^{I}$ real antisymmetric
1316: (therefore no diagonal elements in $H^{I}$) with respect to the
1317: initial and final pair state of an interacting pair. Note that
1318: such a modification is impossible for the single-$j$ model and
1319: therefore is irrelevant for the explanation of the effect at least
1320: for this particular case.
1321:
1322: As shown in Ref. \cite{BFP}, the violation of $T$-invariance in
1323: the form (\ref{4.4}) slightly {\sl enhances} the effect of
1324: predominance of $f_{0}$ rather than reduces it. This might be
1325: understood since the Gaussian ensemble with zero mean averages out
1326: all odd powers of $H^{I}$, eliminating whatever the direct result
1327: of its presence could be and leaving on average only a
1328: renormalization of the real part. Although eq. (\ref{4.4}) keeps
1329: intact the total variance of the random Hamiltonian, the higher
1330: even moments are increased. Thus, if there was a trend (of
1331: different origin) of pushing a state with $J=0$ down, this trend
1332: would be amplified by the inclusion of $H^{I}$.
1333:
1334: Nevertheless, the concept of $T$-invariance may be relevant for
1335: the problem we are interested in. Indeed, any state with $J\neq 0$
1336: appears as a multiplet of degenerate states $|JM\rangle$ with
1337: various projections. For any given $M\neq 0$, the state
1338: $|JM\rangle$ violates the $T$-invariance by choosing the sense of
1339: precession of the angular momentum vector around the quantization
1340: axis. This is nothing but a spontaneous symmetry breaking when the
1341: symmetry of the ground state is lower than that of the
1342: Hamiltonian. As always in such situations, the symmetry is
1343: restored by the degeneracy with other states of the same
1344: multiplet. A physical branch of the excitation spectrum that
1345: restores the symmetry is {\sl rotation} that change the
1346: orientation with no price in energy (Goldstone mode). In this
1347: sense the state with $J=0$ is indeed singled out.
1348:
1349: \subsection{Statistical widths?}
1350:
1351: Many authors, starting with Ref. \cite{BFP}, explored the idea
1352: that the statistical predominance of $J_{0}=0$ states is
1353: associated with the shape of the level density $\rho(E;J)$ for a
1354: given value of $J$. We have already mentioned that two-body
1355: interactions in a finite Hilbert space produce the level density
1356: close to Gaussian in each $J$-class around the centroid of all
1357: strength functions (\ref{3.6}) $F_{k}(E)$ for the basis states
1358: $|k\rangle$ of this class. Then the spectrum has a centroid
1359: \begin{equation}
1360: \bar{E}_{J}=\frac{1}{d_{J}}{\rm Tr}_{J}H\equiv \langle
1361: H\rangle_{J}, \label{4.5}
1362: \end{equation}
1363: and the statistical width
1364: \begin{equation}
1365: \sigma^2_{J}=\langle (H-\langle H\rangle_{J})^{2}\rangle_{J}.
1366: \label{4.6}
1367: \end{equation}
1368: The widths found as a result of statistical spectroscopy
1369: \cite{Kota,Wong} for a given realization, have to be averaged over
1370: the random ensemble.
1371:
1372: It is natural to assume that the $J$-class with a larger
1373: statistical width has a greater chance to contain the ground state
1374: (that does not mean that the inverse statement is also correct).
1375: The results of Ref. \cite{BFP} show that in the $sd$ shell model
1376: for 4 and 6 particles the statistical widths for $J=0$ are by
1377: approximately 10\% larger than for $J=2$ and for other low values
1378: of $J$. Here one needs to notice that, even if this correlation of
1379: $\sigma_{J}$ and $f_{J}$ would be universally correct, we actually
1380: would simply reformulate the original question in another
1381: language, namely what is the reason for the greater statistical
1382: width of the $J=0$ class. But, moreover, the correlation is not
1383: sufficiently strong and not universal \cite{Cov}.
1384:
1385: %Fig. 5
1386: \begin{figure}
1387: \begin{center}
1388: %\vskip 0.5 cm
1389: \includegraphics[width=8 cm]{widthj15N6.eps}
1390: \end{center}
1391: \caption{\label{widthj15}
1392: The average width $\sigma_{J}$ of the
1393: level distribution in the $J$ class as a function of $J$ for 6
1394: particles in a single level $j=15/2$, solid line; predictions of
1395: the statistical formula, dashed line. }
1396: \end{figure}
1397:
1398: %Fig. 3 from Covello book, p. 266
1399: \begin{figure}
1400: \begin{center}
1401: %Fig 6
1402: %\vskip 0.5 cm
1403: \includegraphics[width=8 cm]{wids.eps}
1404: \end{center}
1405: \caption{\label{wids}
1406: Widths $\sigma_{J}$ for various values of
1407: the level spin $j$ in the single-$j$ model as a function of $J$
1408: for 6 particles.}
1409: \end{figure}
1410:
1411: Figs. \ref{widthj15} and \ref{wids} illustrate the situation with
1412: statistical widths in the single-$j$ models. Typically,
1413: $\sigma_{0}$ is greater than the widths of competing low values of
1414: $J$ but the widths $\sigma_{J}$ invariably grow higher than
1415: $\sigma_{0}$ for many large values of $J$. However, only the
1416: fraction of $J_{0}=J_{{\rm max}}$ is noticeably enhanced (still
1417: never on the level more than 15\%) against multiplicity
1418: expectations. One can show that the difference in the widths has
1419: to be much greater, than it is in reality, in order to ensure, at
1420: comparable multiplicities $d_{J}$, the observed predominance of
1421: $J_{0}=0$ states. More subtle effects might be related
1422: \cite{KapPJ} with deviations of higher moments of the statistical
1423: distribution from Gaussian ones, especially for a system of spins
1424: 1/2.
1425:
1426: The comparisons based on pure statistical characteristics of
1427: individual $J$-classes are dangerous since they do not account for
1428: correlations between the classes which are at the core of the
1429: effect. If, for a given realization of the Hamiltonian,
1430: $g_{J}(E)=\rho(E;J)/d_{J}$ is the level density for the $J$-class
1431: normalized to unity, the probability of finding the levels of a
1432: given class at energy higher than $E$ can be written as
1433: \begin{equation}
1434: \chi_{J}(E)= \left(\int_{E}^{\infty} dE'\,g_{J}(E')
1435: \right)^{d_{J}}. \label{4.7}
1436: \end{equation}
1437: Then the probability of having the state $J$ below all other
1438: states will be
1439: \begin{equation}
1440: f_{J}=\int_{-\infty}^{\infty}dE\,\left[-\frac{d}{dE}\chi_{J}(E)\right]
1441: \prod_{J'\neq J}\chi_{J'}(E). \label{4.8}
1442: \end{equation}
1443: The densities of different classes in this formula are strongly
1444: correlated being determined by the same interaction. The task of
1445: averaging this many-point correlation function over the ensemble
1446: of random interactions is hardly solvable.
1447:
1448: \section{Mesoscopic effects of geometry}
1449:
1450: \subsection{General idea}
1451:
1452: To find out how geometry of Hilbert space and of the random
1453: angular momentum coupling can induce the observed effects, we can
1454: first think of even simpler many-body problems \cite{Mul,Yad}.
1455: Take, for instance, a set of $N$ identically interacting spins,
1456: \begin{equation}
1457: H=\frac{1}{2}A\sum_{a\neq b}({\bf s}_{a}\cdot{\bf s}_{b}).
1458: \label{5.1}
1459: \end{equation}
1460: The energy spectrum of this system depends only on the total spin,
1461: ${\bf S}=\sum_{a}{\bf s}_{a}$,
1462: \begin{equation}
1463: E(S)=\frac{A}{2}[S(S+1)-Ns(s+1)], \label{5.2}
1464: \end{equation}
1465: where $s$ is the single-particle spin. If in a random ensemble
1466: the interaction strength
1467: $A$ is symmetrically distributed with respect to zero we
1468: see immediately that the ground state spin will be
1469: either $0$ (antiferromagnetic ordering, $A>0$) or the maximum spin
1470: $S=Ns$ (ferromagnetic ordering, $A<0$), both values appearing with
1471: probability $f_{0}=f_{{\rm max}}=1/2$.
1472:
1473: In this primitive example the answer is simple because the
1474: spectrum of the system is pure rotational and all other quantum
1475: numbers, except for the exact constant of motion, $S$, are not
1476: differentiated by the interaction. We can expect that in all
1477: cases, when the coupling of individual spins plays a role,
1478: rotational modes (as we discussed above, they are Goldstone
1479: excitations restoring the orientational invariance) with the most
1480: ordered spin coupling schemes will bring in an enhanced
1481: probability for the lowest and the highest value of the ground
1482: state spin. The resulting probabilities are decided by the
1483: statistical weights of regions in the parameter space with
1484: positive and negative signs of the moment of inertia. The similar
1485: situation takes place in the case of pairing that is in fact
1486: rotation in gauge space: the energy of a state of seniority $s$ (a
1487: number of unpaired fermions) in the degenerate model for $N$
1488: particles is \cite{Talmi}
1489: \begin{equation}
1490: E(s)=\frac{G}{4}(N-s)(2\Omega-s-N+2), \label{5.3}
1491: \end{equation}
1492: where $\Omega$ is the space capacity. Again, the ground state is
1493: either fully paired one, $s=0$, or ``antipaired" one, $s=N$,
1494: depending on the sign of the pairing constant $G$. Instead of the
1495: rotation operator one has here {\sl quasispin} that is the
1496: generator of $SU(2)$ algebra made of the pair transfer operators
1497: $P,P^{\dagger}$ and particle number operator $N$.
1498:
1499: The situation is very similar in the case of isovector pairing on
1500: a single level. Here the problem can also be solved exactly with
1501: the use of $R(5)$ group formed by six isovector pair creation and
1502: annihilation operators, three components of isospin vector and
1503: total particle number \cite{hecht65a,hecht65b,ginocchio65}. The
1504: Hamiltonian
1505: \begin{equation}
1506: H_{{\rm i.p.}}=V_{0\,1} \, \sum_{\tau=0,\pm 1} P^\dagger_\tau \,
1507: P_{\tau}\,, \label{5.3a}
1508: \end{equation}
1509: has energy eigenvalues
1510: \begin{equation}
1511: E=\frac{V_{0\,1}}{\Omega} \left[
1512: \frac{1}{4}(N-s)(2\Omega-N-s+6)+t(t+1)-T(T+1)\right]\,,
1513: \label{ISO:onel}
1514: \end{equation}
1515: here as before $s$ is a number of unpaired nucleons while $t$
1516: denotes their total isospin.
1517: The presence of the exactly conserved $SU(2)$ subgroup of isospin
1518: makes this example particularly interesting. The correlations
1519: between states with different isospin are represented by the term
1520: $T(T+1)$ which indicates the presence of rotational $V_{0\,1}<0$
1521: or antirotational $V_{0\,1}>0$ bands with 50-50\% probability. In
1522: quadrupole boson models a similar situation occurs often \cite{BF}
1523: with the presence of two ``rotational" Casimir operators,
1524: three-dimensional, $\propto J(J+1)$, and five-dimensional,
1525: $\propto v(v+3)$, where the boson seniority $v$ characterizes the
1526: $O(5)$ group.
1527:
1528: In a general case of complicated fermion dynamics, we can expect
1529: that the geometric chaoticity will on average single out global
1530: rotational modes, so that one can speak about an average
1531: Hamiltonian that describes the relative positions of the {\sl
1532: classes} of states with different exact quantum numbers, such as
1533: total spin and isospin. The same logic works in the case of a
1534: similar problem of the ground state spin in chaotic quantum dots
1535: \cite{Folk}. In our case the effective Hamiltonian $\tilde{H}$ of
1536: classes will take the form of the expansion in powers of the
1537: scalar constant of motion ${\bf J}^{2}$,
1538: \begin{equation}
1539: \tilde{H}=H_{0}+H_{2}{\bf J}^{2}+H_{4}({\bf J}^{2})^{2} +\,...
1540: \label{5.4}
1541: \end{equation}
1542: The coefficients in this expansion are functions of the random
1543: parameters in the original Hamiltonian. In the presence of
1544: additional conserved quantities, as isospin, a similar expansion
1545: in powers of ${\bf T}^{2}$ is to be added to (\ref{5.4}). The
1546: problem is in (approximate) calculation of the coefficients in
1547: this expansion and their averaging over random parameters. If, as
1548: happens in reality, the quadratic term $H_{2}$ dominates, it
1549: determines a rotational band with a random moment of inertia, and
1550: the situation is analogous to that in eq. (\ref{5.2}). For
1551: $J_{{\rm max}}$, the expansion (\ref{5.4}) may not work, and a
1552: special treatment may be needed. This program was first
1553: implemented in Refs. \cite{Mul,oddA,Yad}. A very similar
1554: consideration independently and with different ideology was
1555: carried out for interacting bosons \cite{BF}; we first comment on
1556: the boson problem.
1557:
1558: \subsection{Boson correlations}
1559:
1560: Complicated fermion dynamics generate boson-like collective
1561: excitations. Various types of phonons, magnons and plasmons are
1562: just a few examples for macroscopic systems; shape vibrations and
1563: giant resonances are well studied in mesoscopic physics of nuclei
1564: and atomic clusters. Because only few branches of the spectrum of
1565: elementary excitations are collectivized, the {\sl bosonization}
1566: of the many-fermion problem can lead to significant
1567: simplifications of the formalism and a more transparent physical
1568: picture. This is confirmed by many successes of the IBM
1569: \cite{IBM}, where the boson model is postulated including certain
1570: collective modes although their exact relation to the original
1571: fermion interaction is not rigorously derived. The regular methods
1572: of boson expansion of fermion operators were introduced in nuclear
1573: physics long ago, \cite{BZ1} for a single-$j$ level and \cite{BZ2}
1574: for a general level scheme, see the detailed review article
1575: \cite{MK}. The application of the boson picture to our problem
1576: seems promising, especially because numerous studies of the IBM
1577: with random interactions \cite{BF,Bij,KZC,Kus,KSJ,Zhao2} found a
1578: similar pattern of the predominance of $J=0$ ground states.
1579:
1580: The IBM with two types of interacting bosons has \cite{IBM} the
1581: parameter space sharply divided between the spheres of influence
1582: of different symmetries. For example, this can be illustrated by
1583: the peaks of the ICE \cite{cej} clearly marking the narrow
1584: transitional regions between the symmetries. Using an ansatz of
1585: the axially symmetric coherent intrinsic state generated by a
1586: mixture of bosons, $sp$ for the vibron model and $sd$ for a
1587: conventional nuclear IBM, as a trial function for the ground state
1588: \cite{BF}, one can find the boundaries between the domains of
1589: different symmetry. The coherent state corresponds to the
1590: body-fixed frame with a certain orientation and undetermined spin;
1591: parity is also violated in the vibron intrinsic state. Then one
1592: needs to project out correct angular momentum states (it would be
1593: better but more complicated to minimize the trial energy {\sl
1594: after} this projection) and find the energy spectrum in the
1595: space-fixed frame. A system of $p$-bosons is especially simple
1596: since here quantum numbers $N$ and $J$ determine the state
1597: uniquely (allowed total spins have the same parity as the boson
1598: number and each spin appears once). A similar geometrical pattern
1599: takes place with respect to isospin in odd-odd nuclei \cite{oddA},
1600: where the same statistics is valid for quasideuteron pairs with
1601: $L=0, \,t=1$; as a result some empirical regularities emerge with
1602: random interactions.
1603:
1604: As follows from Ref. \cite{BF}, the different intrinsic shapes of
1605: the ground state, markedly separated in the space of trial
1606: parameters, correspond to different coupling schemes of angular
1607: momentum. Typically one gets, for an even boson number, the
1608: condensate of scalar bosons with the only value $J=0$ possible,
1609: condensate of deformed bosons with the rotational spectrum and the
1610: probability divided between $J_{0}=0$ and $J_{0}=J_{{\rm max}}$
1611: according to the sign of the moment of inertia, and the condensate
1612: of multipole, dipole or quadrupole, quanta again with the same
1613: alternative. Taking into account the corresponding areas in the
1614: parameter space, one comes to a good agreement with ``empirical"
1615: (numerical) data for the random ensemble. One can think of the
1616: used procedure as of a variational method of constructing the
1617: effective Hamiltonian $\tilde{H}$, eq. (\ref{5.4}). In the $sd$
1618: model $\tilde{H}$ includes, apart from the rotational term
1619: $H_{2}J(J+1)$, also an above mentioned term $H'_{2}v(v+3)$. This
1620: picture does not account for the cases with $J_{0}$ or $v_{0}$
1621: different from the edge values but the fraction of such cases is
1622: low. The IBM is however much simpler than fermion systems because
1623: the Bose-statistics creates condensates that in many cases
1624: regularize the angular momentum coupling.
1625:
1626: Going to the fermion system, we can try to use the boson expansion
1627: method. The boson representation of pair operators should be a
1628: reasonable approximation at least for a {\sl dilute} system
1629: \cite{Mul,KZC,KapPJ} with particle number $N$ much smaller than
1630: the space capacity $\Omega$. It follows from the commutator
1631: (\ref{3.22}) (for simplicity we use the single-$j$ model) that
1632: indeed under such conditions the pair operators $P$ and
1633: $P^{\dagger}$ have quasiboson properties:
1634: \begin{equation}
1635: [P_{L'\Lambda'},P^{\dagger}_{L\Lambda}]=\delta_{LL'}
1636: \delta_{\Lambda\Lambda'}+ \; {\rm terms\; of\; order}\;
1637: \frac{N}{\Omega}. \label{5.5}
1638: \end{equation}
1639: Thus, we can introduce the ideal bosons $B_{L\Lambda}$ [$L$ even
1640: from 0 to $L=2j-1=\Omega-2$] and find the boson expansion in
1641: the (symbolically written) form
1642: \begin{equation}
1643: P_{L\Lambda}\approx B_{L\Lambda}+ [B^{\dagger}BB]_{L\Lambda}
1644: +\,\dots. \label{5.6}
1645: \end{equation}
1646:
1647: In the crudest boson approximation, the boson image of the
1648: fermionic Hamiltonian is therefore the gas of $N_{B}=N/2$
1649: noninteracting bosons with quantum numbers $L,\Lambda$ and
1650: energies equal to $V_{L}$,
1651: \begin{equation}
1652: H_{B}^{(0)}=\sum_{L({\rm even})}V_{L}B^{\dagger}_{L}B_{L}.
1653: \label{5.7}
1654: \end{equation}
1655: The ground state of $H_{B}^{(0)}$ is a condensate of all $N$
1656: bosons in a mode with spin $L$ corresponding to the minimum
1657: $V_{L}$. With random choice of $V_{L}$, this leads to a preference
1658: for the ground state spin $J_{0}=0$ \cite{Mul}. Indeed, assume
1659: that every $V_{L}$ has the same chance $1/k$ to be the smallest
1660: one. If the value of $L$, $L({\rm min})$, corresponding to the
1661: smallest $V_{L}$ equals zero, the total spin $J$ can be only zero
1662: as well, the case analogous to the $s$-condensate in the IBM. All
1663: other choices of $L({\rm min})$ create many degenerate states with
1664: energy $E=N_{B}V_{L({\rm min})}$ and various values of total $J$
1665: allowed for a given number of bosons with $L=L({\rm min})$,
1666: including again $J=0$ on more or less equal footing. Summarizing
1667: all cases, we should obtain a $J_{0}=0$ preference. The degeneracy
1668: will be lifted by the interactions coming from higher boson
1669: expansion terms.
1670:
1671: The situation is illustrated by Fig. \ref{N4N5}d, where the
1672: distribution of ground state spins is shown, solid line, for an
1673: ensemble of 5 noninteracting bosons with random energies $V_{L}$
1674: for $L=0,2,4,6,8$ and 10. The dotted line shows that the
1675: statistical distribution of multiplicities is similar to that in
1676: the Fermi-case. The predominance of $f_{0}$ is clearly seen being
1677: however lower here than for fermions. Since the fraction $\sim1/k$
1678: of the pure $s$-condensate falls off for larger systems, we
1679: conclude that the simple bosonic effect does exist but cannot
1680: explain the entire picture. A similar result is seen in Table
1681: \ref{tab1} where the approximation of a 6-fermion system on a
1682: $j=21/2$ level with 3 bosons is shown in column $(b)$.
1683:
1684: %Yours (new) Fig. 4
1685: %Fig 8
1686: \begin{figure}
1687: \begin{center}
1688: \includegraphics[width=11 cm]{N6N7bosons.eps}
1689: \end{center}
1690: \caption{ \label{N6N7}
1691: The distribution of ground state spins for
1692: bosons, $N=6,\,j=4$, left, and $N=7,\,j=6$ in the uniform ensemble
1693: of interaction parameters.}
1694: \end{figure}
1695:
1696: \begin{figure}
1697: \begin{center}
1698: \includegraphics[width=14 cm]{pstat_b.eps}
1699: \end{center}
1700: \caption{ \label{pstatb} The systematics of probabilities $f_0$
1701: for ground state spins $J_0=0$ (circles) and $J_0=J_{\rm max}$
1702: (squares) in various systems of bosons; panels from left to right
1703: show $N=4,6$, and 8 cases as a function of integer single-boson
1704: spin $j$. }
1705: \end{figure}
1706:
1707: We can look also at the real interacting many-boson system
1708: constructed in analogy to our single-$j$ fermionic cases, Figs.
1709: \ref{N6N7} and \ref{pstatb}. The situation here is similar to the
1710: IBM and shows that a considerable probability $f_{J}$ appears only
1711: for $J=0$, $J=J_{max}=Nj$, and $J=j$, which can be explained in
1712: the same way as in the IBM although the number of free parameters
1713: grows with $j$. Despite similarities, the overall boson statistics
1714: has some differences compared to the fermion case, see Figs.
1715: \ref{pstat} and \ref{pstatb}, in particular the probability $f_0$
1716: is generally lower, averaging around 25-35 \%. At the same time
1717: the probability of an ``aligned'' ground state with $J_0=J_{\rm
1718: max}$ is higher and for $N\ll 2j+1$ seem to stabilize at around
1719: 20-25\%.
1720:
1721: \subsection{Sampling method and integrable systems}
1722:
1723: Table 1, column (f), shows the predictions for ground state spins
1724: in a system of $N=6,\,j=21/2$ obtained with the use of the
1725: practical recipe suggested by Zhao, Arima and Yoshinaga
1726: \cite{Zhao1}. The results here are produced by the consecutive
1727: choice of the realizations of the random ensemble with only one
1728: nonzero parameter $V_{L}<0$. In these cases, the absolute value of
1729: this parameter is irrelevant. Thus, one performs the sampling of
1730: the {\sl ``corners"} of the parameter space. For each such extreme
1731: case, the diagonalization determines the ground state spin. The
1732: final prediction for the fraction $f_{J}$ comes as a fraction of
1733: the corners leading to $J_{0}=J$. Although this idea can be used
1734: for fast numerical estimates, it does not shed light on the
1735: physical reasons for the predominance of $J_{0}=0$ just adding
1736: another interesting observation.
1737:
1738: The closest analysis of the recipe suggested in ref. \cite{Zhao1}
1739: shows that the results of this procedure are nor always right,
1740: even qualitatively. A hard problem for such an empirical approach
1741: emerges for the ensembles with different weights for different
1742: matrix elements. The procedure does not indicate how the empirical
1743: fractions obtained for the corners are to be modified for such
1744: cases.
1745:
1746: On the other hand, the corner evaluation indeed works in specific
1747: cases of {\sl integrable} systems. Those are the cases when the
1748: eigenstates of the Hamiltonian can be labeled by exact quantum
1749: numbers as the consequence of rotational and some additional
1750: symmetries. It is known, for example, that for the single level
1751: $j=7/2$ all possible interactions preserve seniority, and the
1752: eigenstates are uniquely characterized by spin and seniority.
1753: Since the eigenfunctions do not depend on the interaction, the
1754: energy eigenvalues are {\sl linear} functions of the interaction
1755: constants \cite{Zhao}, and the coefficients of the linear form are
1756: still determined by geometry of angular momentum coupling. The
1757: search for the set of quantum numbers which provide the minimum of
1758: such a function is reduced to a problem of linear programming. An
1759: elegant geometric method of solving this problem was suggested by
1760: Chau {\sl et al.} \cite{Chau}. It is clear that the knowledge of
1761: corner energies is sufficient for this purpose.
1762:
1763: The situation similar to that for fermions on $j=7/2$ orbit
1764: appears in systems of $p,d$ or $f$ bosons with interaction
1765: conserving the boson number. Again the appropriate quantum numbers
1766: are total spin and boson seniority. The analogous case was
1767: mentioned in the section on boson correlations for the systems
1768: that were not pure in boson composition (for example, $s$ and $p$
1769: or $s$ and $d$ bosons) but were considered with the aid of the
1770: coherent variational function that introduced the condensate of a
1771: specific boson mixture. The result was more complicated because
1772: the proportions of mixture could depend on the interaction that
1773: would make the problem nonlinear. Almost in all cases one again
1774: sees the predominance of the states with extreme values of total
1775: spin or/and seniority.
1776:
1777: \section{Statistical approach}
1778:
1779: \subsection{Effective Hamiltonian}
1780:
1781: In order to come to an estimate of average ground state energy for
1782: randomly interacting fermions we make a simple assumption that
1783: there is a mean field generated by the interactions in each
1784: realization of the ensemble and the field keeps axial symmetry so
1785: that one can speak about mean occupation numbers $n_{m}$ for
1786: fermions in an orbital with $j_{z}=m$ along the symmetry axis
1787: (again we limit ourselves by a formally simpler case of a
1788: single-$j$ level that still keeps the main features of the general
1789: problem). As discussed in Section 2.4, the interaction leads to
1790: equilibration so that the complicated states still can be
1791: characterized by the single-particle occupation numbers. We do not
1792: assume in advance thermal or other specific form of the
1793: equilibrium distribution; instead we can derive it from the
1794: simplest statistical arguments minimizing the ground state energy.
1795: In fact, we just assume that the ground state is as chaotic as
1796: excited states for the majority of realizations. This assumption
1797: is in line with what we had seen in the analysis of the overlaps
1798: of the actual ground state wave functions with fully paired
1799: functions and with the realistic shell model.
1800:
1801: In the mean field approximation, the expectation value of the
1802: Hamiltonian (\ref{3.11}) in a statistical state described by the
1803: occupation numbers $n_{m}$ can be written as
1804: \begin{equation}
1805: E(\{n_{m}\})=\frac{1}{2}\sum_{mm'}V_{mm'}\langle n_{m}n_{m'}
1806: \rangle, \label{6.1}
1807: \end{equation}
1808: where the amplitudes $V_{mm'}$ include the CGC,
1809: \begin{equation}
1810: V_{mm'}= 2\sum_{L\Lambda}V_{L}\left(C^{L\Lambda}_{jm\,jm'}
1811: \right)^{2}. \label{6.2}
1812: \end{equation}
1813: Strictly speaking, eq. (\ref{6.1}) contains the correlated
1814: occupation numbers taken for a given realizations and then
1815: averaged. In the simplest approximation we substitute this by the
1816: product $\langle n_{m}\rangle\langle n_{m'}\rangle$ of mean
1817: occupation numbers; later the sign of averaging will be omitted.
1818:
1819: The occupation numbers are subject to constraints due to the
1820: conservation laws of the particle number $N$ and total angular
1821: momentum projection $M$,
1822: \begin{equation}
1823: \sum_{m}n_{m}=N, \quad \sum_{m}mn_{m}=M. \label{6.3}
1824: \end{equation}
1825: Considering fully aligned states we identify the projection $M$
1826: with total spin $J$. This is similar to what is routinely done in
1827: the nuclear cranking model when applied to the ``rotation" around
1828: the symmetry axis, see for example \cite{Good1}. In this case the
1829: angular momentum is explicitly built up by individual momenta of
1830: the constituents, in keeping with the main idea of geometrical
1831: chaoticity. Thus, we need to minimize the functional
1832: \begin{equation}
1833: \tilde{E}=\frac{1}{2}\sum_{mm'}V_{mm'}n_{m}n_{m'}-\mu\sum_{m}n_{m}
1834: -\gamma\sum_{m}mn_{m}, \label{6.4}
1835: \end{equation}
1836: where we added the constraints (\ref{6.3}) with Lagrange
1837: multipliers of chemical potential, $\mu$, and cranking frequency
1838: (or magnetic field), $\gamma$.
1839:
1840: The extremum of the functional $\tilde{E}(\{n_{m}\})$ is at the
1841: set $\{n_{m}\}$ that satisfies a system of linear algebraic
1842: equations
1843: \begin{equation}
1844: \sum_{m'}V_{mm'}n_{m'}-\mu-\gamma m=0. \label{6.5}
1845: \end{equation}
1846: By solving for $n_{m}(\mu,\gamma)$ and applying the constraints
1847: (\ref{6.3}), we find the appropriate values of the Lagrange
1848: multipliers $\mu(N,M)$ and $\gamma(N,M)$. The kernel (\ref{6.2})
1849: of this inhomogeneous equation contains random parameters and
1850: therefore in general can be inverted. The solution has a form of a
1851: linear polynomial in $m$ (constant occupation for $M=J=0$ and
1852: constant tilt for nonzero $J$). Finally, the value of energy for
1853: this solution is
1854: \begin{equation}
1855: E(N,M)=\tilde{E}(\{n_{m}(\mu(N,M),\gamma(N,M))\}) +\mu(N,M)N
1856: +\gamma(N,M)M. \label{6.6}
1857: \end{equation}
1858: Moreover, since eq. (\ref{6.5}) is still satisfied by this
1859: specific choice of $\mu$ and $\gamma$, it leads to
1860: \begin{equation}
1861: \sum_{mm'}V_{mm'}n_{m}(N,M)n_{m'}(N,M)=\mu(N,M)N+\gamma(N,M)M,
1862: \label{6.7}
1863: \end{equation}
1864: that determines the energy at the extremum in the simple form
1865: \begin{equation}
1866: E(N,M)=\frac{1}{2}[\mu(N,M)N+\gamma(N,M)M] \label{6.8}
1867: \end{equation}
1868: that does not require an explicit expression for the occupation
1869: numbers. Of course, we just applied a standard procedure used in
1870: thermodynamics for the Legendre transformation between different
1871: potentials.
1872:
1873: The value of the chemical potential can be found in a general way
1874: without actually solving the set of equations (\ref{6.5}). Summing
1875: those equations for all $m$ and taking into account that
1876: \begin{equation}
1877: \sum_{m}1=2j+1\equiv \Omega, \quad \sum_{m}m={\rm Tr}\,j_{z}=0,
1878: \label{6.9}
1879: \end{equation}
1880: we obtain
1881: \begin{equation}
1882: \mu=\frac{1}{\Omega}\sum_{mm'}V_{mm'}n_{m'}=\frac{2N}{\Omega^{2}}
1883: \sum_{L}(2L+1)V_{L}, \label{6.10}
1884: \end{equation}
1885: where we used the normalization of the CGC,
1886: \begin{equation}
1887: \sum_{m\Lambda}\left(C^{L\Lambda}_{jm\,jm'}\right)^{2}=\frac{2L+1}
1888: {\Omega}. \label{6.11}
1889: \end{equation}
1890:
1891: In a similar way we can calculate the cranking parameter $\gamma$.
1892: Multiplying eqs. (\ref{6.5}) by $m$ and summing over $m$, we come
1893: to
1894: \begin{equation}
1895: \gamma=\frac{\sum_{mm'}mV_{mm'}n_{m'}}{\sum_{m}m^{2}}=\sum_{L}V_{L}
1896: \sum_{m'}n_{m'}y_{L}(m'), \label{6.12}
1897: \end{equation}
1898: where the geometric factor can be calculated as
1899: \begin{equation}
1900: y_{L}(m')\equiv 2\sum_{m\Lambda}m\left(C^{L\Lambda}_{jm\,jm'}
1901: \right)^{2}=\alpha_{L}m', \label{6.13}
1902: \end{equation}
1903: \begin{equation}
1904: \alpha_{L}=\frac{{\bf L}^{2}-2{\bf j}^{2}}{{\bf L}^{2}}\,
1905: \frac{\sum_\Lambda\Lambda^{2}}{\sum_{m}m^{2}}. \label{6.14}
1906: \end{equation}
1907: Since
1908: \begin{equation}
1909: \sum_{m}m^{2}=\Omega\,\frac{{\bf j}^{2}}{3}, \quad \sum_\Lambda
1910: \Lambda^{2}=(2L+1)\,\frac{{\bf L}^{2}}{3}, \label{6.15}
1911: \end{equation}
1912: the final result reads
1913: \begin{equation}
1914: \gamma=\frac{3}{\Omega^{2}{\bf j}^{4}}\sum_{L}(2L+1)({\bf L}^{2}
1915: -2{\bf j}^{2})\,M. \label{6.16}
1916: \end{equation}
1917: The geometric meaning of the combination in eq. (\ref{6.16}) that
1918: determines the sign of the contribution of pairs with spin $L$ can
1919: be easily understood. By definition (6.4), for $M>0$, the energy
1920: (\ref{6.8}) in the laboratory system goes to minimum for negative
1921: $\gamma$. The inequality ${\bf L}^{2}>2{\bf j}^{2}$ means that the
1922: constituents of the pair are aligned, and this contributes to the
1923: reduction of energy if there is {\sl attraction}, $V_{L}<0$, for
1924: this component of the interaction. Vice versa, there should be
1925: {\sl repulsion}, $V_{L}>0$, for antialigned pairs, ${\bf L}^{2}>
1926: 2{\bf j}^{2}$.
1927:
1928: Identifying $M$ with the magnitude $J$, we come to the effective
1929: Hamiltonian (\ref{5.4}). In the statistical approximation,
1930: $\tilde{H}$ consists of only two terms, $H_{0}$ and $H_{2}$. The
1931: rotational term $H_{2}$ is in this approximation {\sl linear} in
1932: random parameters. Therefore the crude prediction is that the
1933: probability of having the ground state spin $J_{0}=0$ is 1/2.
1934: Although the effective moment of inertia given by the inverse
1935: coefficient in front of $M$ in the parameter $\gamma$, eq.
1936: (\ref{6.16}), does not depend on particle number, the statistical
1937: approach has to work better for a larger $N$. This is indeed seen
1938: in Fig. \ref{pstat} where we juxtapose the results for $f_{J}$ in
1939: the systems of $N=4,6$ and 8.
1940:
1941: One can also note \cite{Yad} that the resulting prediction for the
1942: two items in energy, eq. (\ref{6.8}), picks up the monopole,
1943: $K=0$, and dipole, $K=1$, terms in the Hamiltonian written in the
1944: multipole-multipole form (\ref{3.15}) with an additional factor of
1945: two and the interaction parameters given by the transformation to
1946: the particle-hole channel (\ref{3.17}); the $L$-dependence in the
1947: effective moment inertia (\ref{6.16}) comes from the $6j$-symbol
1948: in eq. (3.17) for $\tilde{V}_{1}$. The extra factor of two
1949: originates from two possible recouplings of single particle
1950: operators in the two-body Hamiltonian on the way to the
1951: statistical approximation (\ref{6.1}). The $K=0$ and $K=1$
1952: multipole interactions are not independent from higher multipoles
1953: [eq. \ref{3.19}]. In the statistical limit the monopole term is
1954: produced in half by $\tilde{V}_0$ in eq. \ref{3.15}, while the
1955: other half comes from all higher $\tilde{V}_K$ combined via eq.
1956: \ref{3.21}, see also discussions and examples in \cite{Volya}.
1957:
1958: The quality of the statistical description can be inferred from
1959: correlations of actual ground state energy in a given copy of the
1960: ensemble with the statistical value, Fig. \ref{esj}. For the
1961: maximum possible momentum the statistical formula works nearly
1962: perfect, for the $J=0$ states the overall correlation is very
1963: good. However, exact energy is shifted down by a constant, which
1964: indicates correlations beyond the statistical description. This
1965: energy shift is shown in Fig. \ref{shift} as a function of the
1966: size of the space $j$ for a six particle system.
1967:
1968: \begin{figure}
1969: \begin{center}
1970: %FIG 2
1971: %\vskip 0.5 cm
1972: \includegraphics[width=14 cm]{esj21N6.eps}
1973: \end{center}
1974: \caption{\label{esj}
1975: For a system of $N=6,\,j=21/2$, individual
1976: realizations with $J_0=0$, left panel, and $J_0=J_{{\rm Max}}$,
1977: right panel, are presented as points with coordinate $x$ equal to
1978: the value of energy from the statistical prediction and
1979: $y$-coordinate corresponding to the exact energy from
1980: diagonalization.}
1981: \end{figure}
1982:
1983: \begin{figure}
1984: \begin{center}
1985: %FIG 2
1986: %\vskip 0.5 cm
1987: \includegraphics[width=7 cm]{shiftN6.eps}
1988: \end{center}
1989: \caption{\label{shift} Average difference in energy between exact
1990: diagonalization and statistical prediction for systems of $6$
1991: particles as a function of $j$.}
1992: \end{figure}
1993:
1994:
1995: \subsection{Occupation numbers}
1996:
1997: The equivalence of the ``monopole + dipole'' truncation to the
1998: results of the lowest statistical approximation means that higher
1999: multipole interactions, being not associated with any conserved
2000: quantities, on average do not influence the equilibrium occupation
2001: numbers. This cannot be always true. Higher multipole interactions
2002: responsible for real deformation of the mean field lift the
2003: remaining degeneracies, compare the bosonic case. However, as $N$
2004: and $\Omega$ grow, many competing multipoles tend to cancel each
2005: other. Therefore we expect the validity of the statistical
2006: approximation to improve as well. Essentially the same result can
2007: be derived by the direct calculation of the cranking model moment
2008: of inertia.
2009:
2010: At the same time, the result above does not mean that the
2011: probability of $J_{0}=J_{{\rm max}}$ also goes to 50\%. The
2012: multiplicity of the class with $J=J_{{\rm max}}$ is very low - in
2013: the single-$j$ model there exists only one such state with the
2014: unique full alignment of all available particles. Although this
2015: state by itself can be described well with the statistical
2016: approach, we cannot reliably compare the energy of this state with
2017: energies of states with other spins split due to the higher
2018: multipole interactions. The fact that the statistical description
2019: works for $J=J_{{\rm max}}$ is seen from Fig. \ref{esj} and Fig.
2020: \ref{N17}, where the average values of the parameters, $\langle
2021: V_{L}\rangle$, obtained from the ensemble copies that resulted in
2022: $J_{0}= J_{{\rm max}}$, are compared to the statistical
2023: predictions.
2024:
2025: %Yours Fig. 7. The way of calculation is unclear.
2026: \begin{figure}
2027: \begin{center}
2028: %FIG 2
2029: %\vskip 0.5 cm
2030: \includegraphics[width=7 cm]{N17.eps}
2031: \end{center}
2032: \caption{\label{N17}
2033: Mean values $\langle V_{L}\rangle$ of
2034: interaction parameters for the cases with $J_{0}=J_{{\rm max}}$ in
2035: the $N=6,\,j=17/2$ system; numerical simulations, solid line, and
2036: statistical predictions, dashed line.}
2037: \end{figure}
2038:
2039: In Ref. \cite{Mul} a stronger statistical assumption was made. The
2040: occupation numbers $n_{m}$ were modeled by those in a Fermi gas
2041: with high temperature when one can neglect dynamical splitting of
2042: single-particle orbitals $|jm)$, and the occupation numbers are
2043: determined solely by the constraints (\ref{6.3}). In this case one
2044: can take
2045: \begin{equation}
2046: n_{m}=\frac{1}{1+\exp(-\alpha-\beta m)}, \label{6.17}
2047: \end{equation}
2048: which ensures the statistical demands, $0\leq n_{m}\leq 1$, and
2049: includes, as in eq. (\ref{6.4}), two parameters associated with
2050: the conservation laws (\ref{6.3}). For any state with $J=M=0$, all
2051: orbitals $|jm)$ must have the same occupancy,
2052: \begin{equation}
2053: n_{m}=\langle 00|a^{\dagger}_{m}a_{m}|00\rangle=
2054: \frac{N}{\Omega}\equiv \bar{n}. \label{6.18}
2055: \end{equation}
2056: $\beta=0$, and the corresponding constant $\alpha_{0}$ is related
2057: to the particle number as
2058: \begin{equation}
2059: \alpha_{0}=\ln\,\frac{\bar{n}}{1-\bar{n}}, \label{6.19}
2060: \end{equation}
2061: vanishing for the half-filled shell when $\bar{n}=1/2$. We usually
2062: try to avoid such systems which are exceptional, see for example
2063: \cite{Zhao1}, because of particle-hole symmetry that eliminates
2064: even-$K$ multipole moments.
2065:
2066: \begin{figure}
2067: \begin{center}
2068: %FIG my 10
2069: %\vskip 0.5 cm
2070: \includegraphics[width=14 cm]{nmall.eps}
2071: \end{center}
2072: \caption{\label{nmall} Average occupation numbers of yrast states
2073: of a given spin $J$ and projection $J_{z}=M=J$ for the
2074: $N=8,\,j=15/2$ system, top row; for $N=6,\,j=17/2$, middle row;
2075: and $N=6,\,j=19/2$, lowest row. Numerical results are shown with
2076: squares with error bars indicating rms deviations; statistical
2077: predictions are indicated with dotted lines.}
2078: \end{figure}
2079:
2080: %\begin{figure}
2081: %\begin{center}
2082: %FIG my 10
2083: %\vskip 0.5 cm
2084: %\includegraphics[width=14 cm]{nm.eps}
2085: %\end{center}
2086: %\caption{\label{nm}
2087: %Average occupation numbers of yrast states of a given spin
2088: %$J$ and projection $J_{z}=M=J$ for the $N=6,\,j=21/2$ system;
2089: %numerical results, solid line, and statistical predictions, dashed
2090: %line.}
2091: %\end{figure}
2092:
2093: For a nonzero total spin projection $M$, the parameter $\beta$,
2094: similar to the cranking frequency $\gamma$, creates a tilt of the
2095: function $n_{m}$. As shown in \cite{Mul}, the main effect is
2096: caught by the expansion of the distribution function (\ref{6.17})
2097: in powers of $\beta$. Due to time-reversal invariance, the
2098: parameter $\alpha$ acquires a correction in the second order, and
2099: the results are
2100: \begin{equation}
2101: \beta=\frac{M}{\bar{n}(1-\bar{n})}\sum m^{2}, \label{6.20}
2102: \end{equation}
2103: \begin{equation}
2104: \alpha=\alpha_{0}+\alpha_{2}, \quad \alpha_{2}=\left(\bar{n}-
2105: \frac{1}{2}\right)\beta^{2}\frac{\sum m^{2}}{\Omega}. \label{6.21_1}
2106: \end{equation}
2107: In accordance with chaotic angular momentum coupling, a nonzero
2108: spin $M$ is created by the fluctuations of occupancies, $\propto
2109: \bar{n}(1-\bar{n})$, as in standard theory of level density of a
2110: Fermi-gas \cite{Bethe,Eric}. The occupancies now can be written as
2111: \begin{equation}
2112: n_{m}=\bar{n}+\frac{mM}{\Omega\langle m^{2}\rangle}-\frac{\bar{n}
2113: -1/2}{\bar{n}(1-\bar{n})}\,\frac{M^{2}}{\Omega^{2}\langle
2114: m^{2}\rangle^{2}}\,(m^{2}-\langle m^{2}\rangle). \label{6.21}
2115: \end{equation}
2116:
2117: This approach works for bosons in the same way. In the case of $N$
2118: bosons on a single $j$ level (integer $j$), the occupation numbers
2119: in the same approximation are given by
2120: \begin{equation}
2121: n_{m}=\frac{1}{\exp(-\alpha-\beta m)}, \label{6.21a}
2122: \end{equation}
2123: and the parameters can be found as
2124: \begin{equation}
2125: \alpha=\alpha_{0}+\alpha_{2}, \quad \alpha_{0}=\ln\,\frac{\bar{n}}
2126: {\bar{n}+1}, \quad \bar{n}=\frac{N}{\Omega}, \label{6.21b}
2127: \end{equation}
2128: \begin{equation}
2129: \beta=\frac{M}{\bar{n}(1+\bar{n})}\sum m^{2}, \label{6.21c}
2130: \end{equation}
2131: \begin{equation}
2132: \alpha_{2}=-\left(\bar{n}+\frac{1}{2}\right)\beta^{2}\frac{\sum
2133: m^{2}}{\Omega}. \label{6.21d}
2134: \end{equation}
2135: The final result is analogous to (\ref{6.21}):
2136: \begin{equation}
2137: n_{m}=\bar{n}+\frac{mM}{\Omega\langle m^{2}\rangle}+\frac{\bar{n}
2138: +1/2}{\bar{n}(1+\bar{n})}\,\frac{M^{2}}{\Omega^{2}\langle
2139: m^{2}\rangle^{2}}\,(m^{2}-\langle m^{2}\rangle), \label{6.21e}
2140: \end{equation}
2141: which predicts the change of curvature compared to the Fermi
2142: expression (\ref{6.21}). Note that the real condensate of bosons
2143: at a single $m$ value is impossible being in contradiction to the
2144: angular momentum requirement.
2145:
2146: The expansion (\ref{6.21}) is rapidly converging because of the
2147: powers of the ``volume" $\Omega$ in the denominator. Using this
2148: for evaluating energies (\ref{6.1}) of the states along the yrast
2149: line \cite{Mul}, one again comes to the effective Hamiltonian in
2150: the form (\ref{5.4}), where the scalar and quadratic terms
2151: coincide with those found in the variational approach. The second
2152: order correction in (\ref{6.21}) adds the quartic term
2153: $H_{4}\propto ({\bf J}^{2})^{2}$. This contribution, which is
2154: small at not very high $J$, was taken in \cite{Mul} to account for
2155: the difference between the observed fraction $f_{0}$ and its
2156: limiting statistical value of 1/2. Such corrections can be
2157: obtained with an improved variational ansatz of the previous
2158: subsection. As seen from Fig. \ref{nmall}, the actual occupation
2159: numbers for given $J=M$, averaged over the ensemble of yrast
2160: states with given $J$, indeed quite well follow the linear
2161: $m$-dependence. There are deviations from the simplest statistical
2162: and variational predictions. The last term in (\ref{6.21})
2163: includes kinematic correlations due to the Fermi-statistics, eq.
2164: \ref{6.17}, however it does not describe fully dynamical effects.
2165: The case of a half-occupied system is particularly interesting:
2166: here the fluctuations are significantly suppressed, see first row
2167: in Fig. \ref{nmall}; accordingly, the term proportional to
2168: $(m^{2}-\langle m^{2}\rangle)$ in eq. (\ref{6.21}) disappears.
2169: However, a regular oscillatory behavior of the occupancies around
2170: the mean statistical behavior survives.
2171:
2172: \subsection{Multipole collectivity}
2173:
2174: We have seen earlier that the structure of the ground state is far
2175: from that of the paired condensate. The question if the set of
2176: random interactions generates, along with the ground states of
2177: zero spin, some collective structure of the excitation spectrum
2178: was put forward already in the original paper \cite{JBD}. To find
2179: out the answer, the authors looked at the saturation of
2180: transitions from the ground state of $J_{0}=0$ to the first
2181: excited state of $J=2$ for a particle-hole operators of quadrupole
2182: type. They observed that it is possible to construct the
2183: quadrupole operator that maximally connects two states, and the
2184: resulting transition accumulates more than 50\% of the
2185: corresponding sum rule, in similarity to the well known
2186: collectivity of the first $2^{+}$ states in non-magic even-even
2187: nuclei. In essence, this emphasizes a particle-hole nature of the
2188: transition with the operator adjusted to each copy of the random
2189: ensemble.
2190:
2191: Actual quadrupole collectivity in nuclei is usually considered
2192: \cite{BM} as a result of coherent interactions in the
2193: particle-hole channel with $K=2$. The background created by
2194: pairing is important since the low-lying phonon collective
2195: excitation should be located within the energy gap due to pairing
2196: \cite{Bel}. In a normal Fermi gas low-lying modes have nearly pure
2197: single-particle character. In contrast to that, in superfluid
2198: systems the presence of the gap stabilizes collective modes as
2199: coherent superpositions of two-quasiparticle excitations. At
2200: sufficiently strong collectivity, the mode found in the random
2201: phase approximation (RPA) becomes unstable, and then effects of
2202: anharmonicity lead to static deformation. In deformed nuclei, the
2203: low-lying quadrupole modes give rise to rotations and new
2204: vibrations around the deformed equilibrium point.
2205:
2206: A comparison of ``normal" quadrupole collectivity with data from
2207: the random interaction ensemble shows \cite{Hor} that collective
2208: effects are strongly suppressed. One needs to "displace" the
2209: ensemble including explicitly a coherent attractive part in order
2210: to reproduce the collectivity \cite{vela,velah}, as was
2211: illustrated long ago by Cortes, Haq and Zuker \cite{cortes}. The
2212: fractional collectivity suggested in Ref. \cite{JBD} was
2213: calculated for 8 particles in the $sd$-shell model for $^{24}$Mg
2214: using the realistic interaction \cite{BW} and various random
2215: ensembles mentioned above, Sect. 4.2. The degree of collectivity
2216: was defined as
2217: \begin{equation}
2218: f.c.=\frac{B({\rm E2}; 0_{1}\rightarrow 2_{1})}{\sum_{n} B({\rm
2219: E2};0_{1}\rightarrow 2_{n})}, \label{6.22}
2220: \end{equation}
2221: where the reduced probability $B({\rm E2})$ of the quadrupole
2222: transition was determined with the {\sl fixed} quadrupole operator
2223: rather than with different operators maximized for each set of
2224: random parameters. This quantity is significantly smaller than
2225: found in Ref. \cite{JBD} for adjusted operators.
2226:
2227: \begin{figure}
2228: \begin{center}
2229: %\vskip 0.5 cm
2230: \includegraphics[width=7 cm]{PRLFIG2.eps}
2231: \end{center}
2232: \caption{ \label{prlfig2}Distribution of
2233: probabilities $B({\rm E2}; 0_{1}\rightarrow
2234: 2_{1})$ in random ensembles $a-d$, Sect. 4.2, in units of the
2235: probability in realistic shell model for $^{24}$Mg \cite{BW}.}
2236: \end{figure}
2237:
2238: Fig. \ref{prlfig2} shows calculated distributions of $B({\rm E2})$
2239: for different ensembles (in units of the transition probability of
2240: 69.5 $e^{2}{\rm fm}^{4}$ found in the $sd$-shell model \cite{BW})
2241: for $^{24}$Mg. Typically, the $B({\rm E2})$ values from random
2242: interactions are by more than an order of magnitude weaker than in
2243: realistic calculations. Even the maximum $B({\rm E2})$ values out
2244: of 1000 samples for all four models are smaller than the realistic
2245: value, although is few copies they come close. The distribution of
2246: the $B({\rm E2})$ values for models (a) and (b) is close to the
2247: Porter-Thomas (\ref{2.6}) as expected for matrix elements of a
2248: simple operator between two chaotic states \cite{Brody,SF,ann};
2249: the first excited state is even less regular than the ground
2250: state. The model (c) with realistic pairing generates a hint of
2251: collectivity. This agrees with what we have said above concerning
2252: the role of pairing correlations supporting the multipole
2253: collectivity. The sharp cutoff at small values of $B({\rm E2})$ in
2254: model (d) happens close to the value that can be obtained for the
2255: pure $(d_{5/2})_{p}^{4}(d_{5/2})_{n}^{4}$ configuration since here
2256: the multipole-multipole correlations generated by the higher-$L$
2257: components of interaction are absent.
2258:
2259: We have seen, Fig. \ref{rpairing}, that, in spite of mainly
2260: chaotic nature of eigenstates for random interactions, there
2261: exists a slight excess of cases with a significant overlap between
2262: the ground state wave function and the fully paired state. This
2263: excess is particularly noticeable in cases when the sequence of
2264: the lowest states is $J_{0}=0,\,J_{1}=2$ (such cases appear also
2265: with a higher probability than could be expected from the
2266: statistical multiplicity). This means that there is a probability,
2267: exceeding the expectation of pure random models, that random
2268: interactions indeed create collective effects. To illustrate this
2269: point, we consider the dynamical quantity, which we call {\sl
2270: Alaga ratio}, that can distinguish between different collective
2271: structures,
2272: \begin{equation}
2273: A=\frac{{Q}^2}{B(\rm E2)}. \label{6.22a}
2274: \end{equation}
2275: Here the numerator is the expectation value squared of the
2276: quadrupole moment of the first excited state $2_{1}$,
2277: \begin{equation}
2278: {Q}=\langle J M=J|{M}_{2 0}|J\, M=J\rangle\, \label{6.23}
2279: \end{equation}
2280: and the $B({\rm E2})$ transition strength in the denominator is
2281: defined as
2282: \begin{equation}
2283: B{\rm (E2)}=\sum_{M_f \kappa}\,|\langle J M_f|{M}_{2 \kappa}|J\,
2284: M_i \rangle |^2\,. \label{6.24}
2285: \end{equation}
2286: Thus, limiting ourselves by the sequences $J_{0}=0,\,J_{1}=2$, we
2287: are looking at the ratio of diagonal to off-diagonal matrix
2288: elements of the quadrupole operator.
2289:
2290: If random coupling of individual spins results in average
2291: spherical shape, and the ground and the first excited state are of
2292: similar structure, the Alaga ratio (\ref{6.22a}) should be small.
2293: On the other hand, if this random coupling creates a more or less
2294: rigid structure, one can expect the fulfillment of Alaga intensity
2295: rules \cite{BM},
2296: \begin{equation}
2297: A=\frac{4}{49}\equiv A_{0}\,. \label{6.25}
2298: \end{equation}
2299: For the two lowest states without any genetic interconnection, the
2300: Alaga ratio can take any value. In the case of single $j=15/2$
2301: model, the sequences of interest appear in 6.7\% for $N=4$ and in
2302: 9.2\% cases for $N=6$. The distribution of the Alaga ratio for
2303: these systems, Fig. \ref{lalaga}, reveals two peaks at $A=0$ and
2304: $A=A_{0}$. The peaks are pronounced stronger at a larger particle
2305: number. The idea that such structure can arise from random
2306: interactions is not that surprising. It was proven long ago
2307: \cite{Rock} that the effects of interactions in large
2308: non-superfluid rotating Fermi-systems cancel leaving the
2309: rigid-body moment of inertia. In agreement with this, the
2310: statistical dependence of the level density, based on the
2311: geometrical chaoticity, also gives the same value of the moment of
2312: inertia, see \cite{Eric} and eq. (\ref{3.5}). The Alaga ratio
2313: seems to be a more sensitive signature of rotational behavior than
2314: the standard ratio of energies involving the next $J=4$ state.
2315:
2316: \begin{figure}
2317: \begin{center}
2318: %\vskip 0.5 cm
2319: \includegraphics[width=14 cm]{lalaga.eps}
2320: \end{center}
2321: \caption{ \label{lalaga} The Alaga ratio for 6 and 4 particles in
2322: a single $j=15/2$ level; the histogram includes the states with
2323: the sequence $J_{0}=0, J_{1}=2$ of lowest spins.}
2324: \end{figure}
2325:
2326: \subsection{Multipole dynamics}
2327:
2328: The multipole dynamics can be studied analytically starting with
2329: the operator equations of motion for the multipole operators
2330: $M_{K\kappa}$ and using the Hamiltonian in a suitable form
2331: (\ref{3.15}) and commutation relations (\ref{3.24}). The exact
2332: equations of motion are
2333: \[[M_{K\kappa},H]=-g_{K}\sum_{K'S}g_{K'}g_{S}\tilde{V}_{K'}
2334: [1-(-)^{K+K'+S}]\left\{\begin{array}{ccc}
2335: K & K' & S\\
2336: j & j & j\end{array}\right\}\]
2337: \begin{equation}
2338: \times\sum_{\kappa'\sigma}(-)^{\sigma}\left(\begin{array}{ccc}
2339: K & K' & S\\
2340: \kappa & \kappa' & -\sigma\end{array}\right)\frac{1}{2}
2341: [M_{S\sigma},M^{\dagger}_{K'\kappa'}]_{+}, \label{6.26}
2342: \end{equation}
2343: where $[...\,,\,...]_{+}$ denotes an anticommutator, and
2344: $g_{K}=\sqrt{2K+1}$. Note that zero values $K'$ and $S$ do not
2345: contribute to these equations, and $K=0$ or 1 give trivial zero
2346: results because of the conservation laws.
2347:
2348: We take in eqs. (\ref{6.26}) the matrix element between the ground
2349: state $|0\rangle$, assumed to have zero spin, and a hypothetical
2350: collective state $|K\kappa\rangle$. After separating the dependence on
2351: magnetic quantum numbers by the Wigner-Eckart theorem,
2352: \begin{equation}
2353: \langle S\sigma|M_{K'\kappa'}|K\kappa\rangle=(-)^{\sigma+\kappa'}
2354: \left(\begin{array}{ccc} S & K' & K \\
2355: -\sigma & -\kappa' & \kappa\end{array}\right)M^{K'}_{SK},
2356: \label{6.27}
2357: \end{equation}
2358: we come to the set of nonlinear equations for the matrix elements
2359: \begin{equation}
2360: M_{K}\equiv \langle 0|M_{K\kappa}|K\kappa\rangle\equiv (-)^{K}
2361: \frac{1}{g_{K}}\,M^{K}_{0K}, \label{6.28}
2362: \end{equation}
2363: that contain the excitation energy $E_{K}$ of the collective
2364: state,
2365: \begin{equation}
2366: E_{K}g_{K}M_{K}=\sum_{K'S}g_{K'}g_{S}
2367: (\tilde{V}_{K'}-\tilde{V}_{S})\frac{1-(-)^{K+K'+S}}{2} \left\{
2368: \begin{array}{ccc}
2369: K & K' & S \\
2370: j & j& j \end{array}\right\}M_{K'}M^{S}_{K'K}. \label{6.29}
2371: \end{equation}
2372: Those equations are still exact if the sum runs over all allowed
2373: intermediate states of spin $K'$.
2374:
2375: In the spirit of the RPA we make here truncation leaving only the
2376: most coherent contributions with $K'=K$ when the multipole $K$
2377: does not share its angular momentum with other excitations.
2378: Thereby the equation gets linearized, and we obtain a closed
2379: expression for the excitation energy
2380: \begin{equation}
2381: E_{K}=\sum_{S({\rm odd})}g_{S} (\tilde{V}_{K}-\tilde{V}_{S})
2382: \left\{\begin{array}{ccc}
2383: K & K & S \\
2384: j & j& j \end{array}\right\}M^{S}_{KK}. \label{6.30}
2385: \end{equation}
2386: The result (\ref{6.30}) depends on expectation values $M^{S}_{KK}$
2387: of odd-spin multipoles; the whole dynamics is concentrated in the
2388: differences $\tilde{V}_{K}-\tilde{V}_{S}$ of multipole coupling
2389: constants. The diagonal matrix elements of {\sl odd} multipoles of
2390: rank $S$ can be constructed as those of irreducible tensors made
2391: of the $S$ components of angular momentum ${\bf J}$, another
2392: manifestation of fractional parentage or quasi-random geometrical
2393: coupling.
2394:
2395: Using again the commutators (\ref{3.24}) for odd $K'$ and applying
2396: the same RPA-like approximation, we come to
2397: \begin{equation}
2398: M^{S}_{KK}=2g_{K}^{2}g_{S}\left\{\begin{array}{ccc} K & S &K\\
2399: j & j & j\end{array}\right\}, \quad S =\,{\rm odd}. \label{6.31}
2400: \end{equation}
2401: In the semiclassical limit of large $j$ the $6j$-symbol in eq.
2402: (\ref{6.31}) is proportional to the Legendre polynomial
2403: $P_{S}(\cos\theta)$, where $\theta$ is the angle between ${\bf j}$
2404: and ${\bf K}$. This describes the reorientation of the occupation
2405: numbers in the process of collective excitation by the multipole
2406: $K$. In this approximation, the collective excitation energy is
2407: given by
2408: \begin{equation}
2409: E_{K}=2g_{K}\sum_{S({\rm odd})}g_{S}^{2}
2410: (\tilde{V}_{K}-\tilde{V}_{S}) \left\{\begin{array}{ccc}
2411: K & K & S \\
2412: j & j& j \end{array}\right\}^{2}. \label{6.32}
2413: \end{equation}
2414: The earlier discussed statistical limit, $E_{K}\approx
2415: \tilde{V}_{1}K(K+1)$, is given by the contribution of the $S=1$
2416: term. The other terms determine the collectivity corrections.
2417:
2418: \begin{figure}
2419: \begin{center}
2420: %FIG my 10
2421: %\vskip 0.5 cm
2422: \includegraphics[width=7 cm]{mrj15n6g.eps}
2423: \end{center}
2424: \caption{\label{nrj15n6} Transition energies between the ground
2425: state $J_{0}=0$ and the yrast state $J=K$ for a system
2426: $N=6\,j=15/2$ in a random ensemble compared to the RPA
2427: approximation, eq. (\ref{6.32}).}
2428: \end{figure}
2429:
2430: Fig. 15 shows the correspondence between the excitation energies
2431: found in many runs with random interactions for the $N=6,\,j=15/2$
2432: system and the RPA prediction (\ref{6.32}). The strong correlation
2433: observed here confirms that, even for random interactions, the
2434: geometry of space and spin coupling forms coherent combinations
2435: that can be assessed with the regular methods of many-body theory.
2436: Also a horizontal line of point accumulation is seen that
2437: corresponds to the situation when the low-lying states are nearly
2438: degenerate. It would be interesting to study what part of the
2439: random parameter space corresponds to such absence of collective
2440: correlations.
2441:
2442: \section{Conclusion}
2443:
2444: The idea to study the physics of a finite mesoscopic system with
2445: random interactions turned out to be very fruitful. Many
2446: theoreticians responded to the challenge; many new things were
2447: learned even for the simplest systems of few identical particles;
2448: many questions are not answered yet. After four years of extensive
2449: studies the main result can be formulated as following: standard
2450: textbook ideas of the factors that form the low-lying structure of
2451: a closed self-sustaining mesoscopic systems are insufficient. The
2452: quantum numbers of the ground states and some regularities of
2453: spectra emerge not necessarily due to the corresponding coherent
2454: parts of the interparticle interaction. Up to a large extent,
2455: these characteristics are predetermined by the conservation laws
2456: and geometry of available single-particle space.
2457:
2458: The angular moment $J_{0}=0$ of the ground state in even-even
2459: nuclei appears with a probability of the order 50\% with any
2460: randomly taken rotationally invariant residual interaction. The
2461: underlying mechanism may be related to chaotic coupling of
2462: individual spins that creates an average yrast line described by
2463: the effective Hamiltonian with mainly quadratic dependence on
2464: total angular momentum. The time-reversible $J=0$ state turns out
2465: to have an exceptional ability of coming at the bottom or at the
2466: top end of the average spectrum. The strong attractive $L=0$
2467: pairing of identical particles amplifies this effect and leads in
2468: reality to a 100\% probability of $J_{0}=0$. Similar regularities
2469: are associated with odd-$A$ and odd-odd systems. We relied mostly
2470: on the nuclear structure as the basic and the best studied object
2471: of applications. However, the similar physics of random
2472: interactions and chaotic spin coupling certainly plays a role in
2473: other mesoscopic systems, such as atomic clusters, metallic
2474: grains, quantum dots and quantum spin glasses.
2475:
2476: The new avenue were opened for general studies of quantum chaos.
2477: From extremes of random matrix theory based on the most general
2478: canonical Gaussian ensembles and later on the two-body random
2479: ensemble, we proceed to investigation of random interactions fully
2480: compatible with exact symmetries of finite systems. The new
2481: effects that enter the game now are correlations between the
2482: classes of states of different symmetry being governed by the
2483: common Hamiltonian with a relatively small number of random
2484: parameters. These correlations bring in the traces of new order
2485: and collectivity solely created by random interactions along with
2486: exact symmetries.
2487:
2488: The pioneering paper \cite{JBD} on the subject was concluded by
2489: the words that their studies ``have barely scratched the surface
2490: of possible questions". Now we know more and perhaps can say that
2491: we started to penetrate that surface. Nevertheless the problem is
2492: not entirely solved, and the future advances seem to be quite
2493: exciting and promising.
2494:
2495: \section{Acknowledgments}
2496:
2497: The authors are indebted to B.A. Brown, M. Horoi, D. Mulhall and
2498: J. Roebke for creative collaboration. Constructive discussions
2499: with G.F. Bertsch, R. Bijker, P. Cejnar, D. Dean, V.V. Flambaum,
2500: A, Frank, F.M. Izrailev, D. Kusnezov, T. Papenbrock, N.A.
2501: Smirnova, O.P. Sushkov, P. Van Isacker, H.A. Weidenm\"{u}ller,
2502: Y.M. Zhao and A.P. Zuker at various stages of the work are
2503: gratefully acknowledged. We acknowledge support from the NSF,
2504: Grants Nos. PHY-0070911 and PHY-0244453, and from the US
2505: Department of Energy, Nuclear Physics Division, under Contract No.
2506: W-31-109-ENG-38.
2507:
2508: \begin{thebibliography}{99}
2509: \bibitem{Racah} G. Racah, Phys. Rev. 62 (1942) 438; 63 (1943) 367.
2510: \bibitem{Talmi} I. Talmi, {\sl Simple Models of Complex Nuclei}
2511: (Harwood, New York, 1993).
2512: \bibitem{MJ} M.G. Mayer and J.H.D. Jensen, {\sl Elementary Theory
2513: of Nuclear Shell Structure} (Wiley, New York, 1955).
2514: \bibitem{BMP} A. Bohr, B.R. Mottelson and D. Pines, Phys. Rev.
2515: 110 (1958) 936.
2516: \bibitem{Bel} S.T. Belyaev, Mat. Fys. Medd. Dan. Vid. Selsk. 31
2517: (1959) No. 11.
2518: \bibitem{KS} L.S. Kisslinger and R.A. Sorensen, Mat. Fys. Medd.
2519: Dan. Vid. Selsk. 32 (1960) No. 9.
2520: \bibitem{BAB} B.A. Brown, Prog. Part. Nucl. Phys. 47 (2001) 517.
2521: \bibitem{JBD} C.W. Johnson, G.F. Bertsch and D.J. Dean, Phys.
2522: Rev. Lett. 80 (1998) 2749.
2523: \bibitem{grib} V.V. Flambaum, A.A. Gribakina, G.F. Gribakin and
2524: M.G. Kozlov, Phys. Rev. A 50 (1994) 267.
2525: \bibitem{big} V. Zelevinsky, B.A. Brown, N. Frazier and M. Horoi,
2526: Phys. Rep. 276 (1996) 85.
2527: \bibitem{ann} V. Zelevinsky, Annu. Rev. Nucl. Part. Sci. 46 (1996)
2528: 237.
2529: \bibitem{Fluc} {\sl Statistical Theories of Spectra:
2530: Fluctuations}, ed. C.E. Porter (Academic Press, New York, 1965).
2531: \bibitem{Brody} T.A. Brody, J. Flores, J.B. French, P.A. Mello, A.
2532: Pandey and S.S.M. Wong, Rev. Mod. Phys. 53 (1981) 385.
2533: \bibitem{Guhr} T. Guhr, A. M\"{u}ller-Groeling, and H.A.
2534: Weidenm\"{u}ller, Phys. Rep. 299 (1998) 189.
2535: \bibitem{TBRE} J.B. French and S.S.M. Wong, Phys. Lett. B 33
2536: (1970) 449.
2537: \bibitem{TBREa} O. Bohigas and J. Flores, Phys. Lett. B 34 (1971)
2538: 261.
2539: \bibitem{Mehta} M.L. Mehta, {\sl Random Matrices} (Academic Press,
2540: Boston, 1991).
2541: \bibitem{Kota} V.K.B. Kota, Phys. Rep. 347 (2001) 223.
2542: \bibitem{embe} J.B. French, Rev. Mex. Fis. 22 (1973) 221.
2543: \bibitem{mon} K.K. Mon and J.B. French, Ann. Phys. (N.Y.) 95
2544: (1975) 90.
2545: \bibitem{mf} V.G. Zelevinsky, Nucl. Phys. A555 (1993) 109.
2546: \bibitem{izr} F.M. Izrailev, Phys. Rep. 196 (1990) 299.
2547: \bibitem{entr} V. Zelevinsky, M. Horoi and B.A. Brown, Phys. Lett.
2548: B350 (1995) 141.
2549: \bibitem{temp} M. Horoi, V. Zelevinsky and B.A. Brown, Phys. Rev.
2550: Lett. 74 (1995) 5194.
2551: \bibitem{perc} I.C. Percival, J. Phys. B 6 (1973) L229.
2552: \bibitem{verbaar} J.J.M. Verbaarschot and P.J. Brussard, Phys.
2553: Lett. B87 (1979) 155.
2554: \bibitem{kilg} G. Kilgus {\sl et al.}, Z. Phys. A326 (1987) 41.
2555: \bibitem{SF} O.P. Sushkov and V.V. Flambaum, JETP Lett. 32 (1980)
2556: 353; Sov. Phys. Usp. 25 (1982) 1.
2557: \bibitem{AB} N. Auerbach and B.A. Brown, Phys. Lett. B340 (1994)
2558: 6.
2559: \bibitem{Alfim} V.P. Alfimenkov {\sl et al.}, Nucl. Phys. A398
2560: (1983) 93.
2561: \bibitem{Bowm} J.D. Bowman {\sl et al.}, Phys. Rev. Lett. 65
2562: (1990) 1192; C.M. Frankle {\sl et al.}, Phys. Rev. Lett. 67 (1991)
2563: 564. \bibitem{Mitch} G.E.Mitchell, J.D.Bowman and
2564: H.A.Weidenmüller, Rev. Mod. Phys. 71 (1999) 445.
2565: \bibitem{Dan} G.V. Danilyan {\sl et al.}, JETP Lett. 26 (1977)
2566: 186.
2567: \bibitem{Petr} G.A. Petrov, Nucl. Phys. A502 (1989) 297.
2568: \bibitem{ZVYad} V. Zelevinsky and A. Volya, Yad. Fiz. 66 (2003)
2569: 1829 [Phys. At. Nucl. 66 (2003) 1781].
2570: \bibitem{FIC} V.V. Flambaum, F.M. Izrailev and G. Casati, Phys.
2571: Rev. E 54 (1996) 2136.
2572: \bibitem{FI} V.V. Flambaum and F.M. Izrailev, Phys. Rev. E 55,
2573: (1997) R13; E 56 (1997) 5144.
2574: \bibitem{mark} M. Srednicki, Phys. Rev. E 50 (1994) 888.
2575: \bibitem{Sok} V.V. Sokolov, B.A. Brown and V. Zelevinsky, Phys.
2576: Rev. E 58 (1998) 56.
2577: \bibitem{cej} P. Cejnar, V. Zelevinsky and V.V. Sokolov, Phys.
2578: Rev. E 63 (2001) 036127.
2579: \bibitem{ICE} A. Volya and V. Zelevinsky, Phys. Lett. B, in press
2580: (2003).
2581: \bibitem{Peres} A. Peres, {\sl Quantum theory: Concepts and
2582: Methods} (Kluwer, Dordrecht, 1995).
2583: \bibitem{Niels} M.A. Nielsen and I.L. Chuang, {\sl Quantum
2584: Computation ands Quantum Information} (Cambridge University Press,
2585: 2000).
2586: \bibitem{Pros} T. Prosen and M. \u{Z}nidari\u{c}, J. Phys. A 35
2587: (2002) 1455.
2588: \bibitem{boh} O. Bohigas, M.J. Giannoni and C. Schmit, Phys. Rev.
2589: Lett. 52 (1984) 1.
2590: \bibitem{GuP} I.I. Gurevich and M.I. Pevzner, Nucl. Phys. 2
2591: (1957) 575.
2592: \bibitem{HRM} H.L. Harney, A. Richter and H.A. Weidenm\"{u}ller,
2593: Rev. Mod. Phys. 58 (1986) 607.
2594: \bibitem{Doss} T. D\o ssing {\sl et al.}, Phys. Rep. 268 (1996) 1.
2595: \bibitem{Bethe} H. Bethe, Phys. Rev. 50 (1936) 332.
2596: \bibitem{Eric} T. Ericson, Adv. Phys. 9 (1960) 425.
2597: \bibitem{Wig} E.P. Wigner, Am. J. Math. 63 (1941) 57.
2598: \bibitem{regge} G. Ponzano and T. Regge, in {\sl Spectroscopic and
2599: Group Theoretical Methods in Physics} (North Holland, Amsterdam,
2600: 1968) p. 1.
2601: \bibitem{Bieden} L. Biedenharn and J.D. Louck, {\sl Angular Momentum
2602: in Quantum Physics} (Addison-Wesley, Reading, 1981).
2603: \bibitem{Wong} S.S.M. Wong, {\sl Nuclear Statistical Spectroscopy}
2604: (Oxford University Press, New York, 1986).
2605: \bibitem{Fraz} N. Frazier, B.A. Brown and V. Zelevinsky, Phys.
2606: Rev. C 54 (1996) 1665.
2607: \bibitem{good} A.L. Goodman, Phys. Rev. C 58 (1998) R3051.
2608: \bibitem{Volya} A. Volya, Phys. Rev. C 65 (2002) 044311.
2609: \bibitem{EP} A. Volya, B.A. Brown and V. Zelevinsky, Phys. Lett. B
2610: 509 (2001) 37.
2611: \bibitem{JBDT} C.W. Johnson, G.F. Bertsch, D. J. Dean and I.
2612: Talmi, Phys. Rev. C 61 (1999) 014311.
2613: \bibitem{BFP} R. Bijker, A. Frank, and S. Pittel. Phys. Rev. C 60
2614: (1999) 021302.
2615: \bibitem{BF} R. Bijker and A. Frank, Phys. Rev. Lett. 84 (2000)
2616: 420; Phys. Rev. C 62 (2000) 014303; Phys. Rev. Lett. 87 (2001)
2617: 029201; Phys. Rev. C 64 (2001) 061303; Phys. Rev. C 65 (2002)
2618: 044316.
2619: \bibitem{Bij} R. Bijker, nucl-th/0303069,
2620: \bibitem{KZC} D. Kusnezov, N.V. Zamfir and R.F. Casten, Phys. Rev.
2621: Lett. 85 (2000) 1396.
2622: \bibitem{Kus} D. Kusnezov, Phys. Rev. Lett. 85 (2000) 3773; 87
2623: (2001) 029202.
2624: \bibitem{KSJ} L.F. Santos, D. Kusnezov and P. Jacquod, Phys. Lett.
2625: B 537 (2002) 62.
2626: \bibitem{Mul} D. Mulhall, A. Volya and V. Zelevinsky. Phys. Rev.
2627: Lett. 85 (2000) 4016.
2628: \bibitem{Hor} M. Horoi, B.A. Brown and V. Zelevinsky, Phys. Rev.
2629: Lett. 87 (2001) 062501.
2630: \bibitem{oddA} M. Horoi, A. Volya and V. Zelevinsky, Phys. Rev. C
2631: 66 (2002) 024319.
2632: \bibitem{Mul1} D. Mulhall, A. Volya and V. Zelevinsky, Nucl. Phys.
2633: A 682 (2001) 229c.
2634: \bibitem{Cov} V. Zelevinsky and A. Volya, in {\sl Challenges of
2635: Nuclear Structure}, ed. A. Covello (World Scientific, Singapore,
2636: 2002) p. 261.
2637: \bibitem{Yad} V. Zelevinsky, D. Mulhall and A. Volya, Phys. Atom.
2638: Nucl. 64 (2001) 525.
2639: \bibitem{Mul2} D. Mulhall, V. Zelevinsky and A. Volya, Acta Phys.
2640: Polonica B 32 (2001) 2491.
2641: \bibitem{Zhao} Y.M. Zhao and A. Arima, Phys. Rev. C 64 (2001)
2642: 041301.
2643: \bibitem{Zhao1} Y.M. Zhao, A. Arima and N. Yoshinaga, Phys. Rev. C
2644: 66 (2002) 034302.
2645: \bibitem{Zhao2} Y.M. Zhao, A. Arima and N. Yoshinaga, Phys. Rev. C
2646: 66 (2002) 064322, 064323.
2647: \bibitem{Arima} A. Arima, N. Yoshinaga and Y.M. Zhao, Eur. Phys. J.
2648: A 13 (2002) 105,
2649: \bibitem{Yoshi} N. Yoshinaga, A. Arima and Y.M. Zhao, J. Phys. A 35
2650: (2002) 8575.
2651: \bibitem{Zhao3} Y.M. Zhao, A. Arima and N. Yoshinaga, Phys. Rev. C 68
2652: (2003) 014322.
2653: \bibitem{Droz} S. Drozdz and M. Wojcik. Physika A 301 (2001) 291.
2654: \bibitem{vela} V. Vel\'{a}zquez and A.P. Zuker, Phys. Rev. Lett.
2655: 88 (2002) 072502.
2656: \bibitem{Zhaocomm} Y.M. Zhao and A. Arima, nucl-th/0202056.
2657: \bibitem{velah} V. Vel\'{a}zquez, J.G. Hirsch, A. Frank and A.P.
2658: Zuker, Phys. Rev. C 67 (2003) 034311.
2659: \bibitem{KAPP} L. Kaplan and T. Papenbrock, Phys. Rev. Lett. 84
2660: (2000) 4553.
2661: \bibitem{KapPJ} L. Kaplan, T. Papenbrock and C.W. Johnson, Phys.
2662: Rev. C 63 (2001) 014307.
2663: \bibitem{Chau} P. Chau Huu-Tai, A. Frank, N.A. Smirnova and
2664: P. Van Isacker, Phys. Rev. C 66 (2002) 061302.
2665: \bibitem{rowe} G. Rosensteel and D. J. Rowe Phys. Rev. C 67
2666: (2003) 014303.
2667: \bibitem{dcs} A.A. Anselm, Phys. Lett. B 217 (1989) 169.
2668: \bibitem{ocs} A. Volya, S. Pratt and V. Zelevinsky. Nucl. Phys.
2669: A671 (2000) 617.
2670: \bibitem{BW} B.A. Brown and B.H. Wildenthal, Annu. Rev. Nucl.
2671: Part. Sci. 38 (1988) 29.
2672: \bibitem{Sush} J. Oitmaa and O.P. Sushkov, Phys. Rev. Lett. 87
2673: (2001) 167206.
2674: \bibitem{hecht65a} K. T. Hecht, Nucl. Phys. 63 (1965) 177.
2675: \bibitem{hecht65b} K. T. Hecht, Phys. Rev. 139 B (1965) 794.
2676: \bibitem{ginocchio65} J. N. Ginocchio, Nucl. Phys. 74 (1965)
2677: 321.
2678: \bibitem{Folk} J.A. Folk, C.M. Markus, R. Berkovits, I.L. Kurland,
2679: I.L. Aleiner and B.L. Altshuler, Phys. Scr. T90 (2001) 26.
2680: \bibitem{IBM} F. Iachello and A. Arima, {\sl The Interacting Boson
2681: Model} (Cambridge University Press, 1987).
2682: \bibitem{BZ1} S.T. Belyaev and V.G. Zelevinsky, Sov. Phys. JETP 15
2683: (1962) 1004.
2684: \bibitem{BZ2} S.T. Belyaev and V.G. Zelevinsky, Nucl. Phys. 39
2685: (1962) 582.
2686: \bibitem{MK} A. Klein and E.R. Marshalek, Rev. Mod. Phys. 63
2687: (1991) 375.
2688: \bibitem{Good1} A.L. Goodman, Phys. Rev. Lett. 73 (1994) 416,
2689: 1734.
2690: \bibitem{BM} A. Bohr and B. Mottelson, {\sl Nuclear Structure}
2691: (Benjamin, N.Y., 1974) vol. 2.
2692: \bibitem{cortes} A. Cortes, R.U. Haq and A.P. Zuker, Phys. Lett.
2693: 115B (1982) 1.
2694: \bibitem{Rock} R.M. Rockmore, Phys. Rev. 124 (1961) 27.
2695: %not sorted references
2696:
2697: \end{thebibliography}
2698: \end{document}
2699:
2700: